Program Book - Artificial Intelligence Association of Thailand (AIAT)

Transcription

Program Book
The 14th Pacific Rim International Conference on
Artificial Intelligence
(PRICAI 2016)
The 19th International Conference on
Principles and Practice of Multi-Agent Systems
(PRIMA 2016)
The 2016 Pacific Rim Knowledge Acquisition Workshop
(PKAW 2016)
“Imagination is more
important than knowledge.”
- Albert Einstein
PRICAI/PRIMA/PKAW 2016 Program Book
About this publication
Title: The Program Book of the 14th Pacific Rim International
Conference on Artificial Intelligence (PRICAI 2016), and the 19th International Conference on Principles and Practice of Multi-Agent Systems (PRIMA 2016), August 22- 26, 2016, Phuket, Thailand
Editor-in-chief:
Thanaruk Theeramunkong
Editors: Matteo Baldoni, Richard Booth, Katsutoshi Hirayama,
Narit Hnoohom, Chuleerat Jaruskulchai, Mahasak Ketcham,
Sanparith Marukatat, Masayuki Numao, Hayato Ohwada, Manabu Okumura, Patiyuth Pramkeaw,
Merlin Teodosia Suarez, Thepchai Supnithi, Paolo Torroni, Kenichi Yoshida, Min-Ling Zhang
Production Assistants: Thodsaporn Chay-intr, Uraiwan Buatoom,
Rachasak Somyanonthanakul, Choermath Hoongakkaraphan
Cover Designer:
Winnaphat Phatcharaphiphattanakorn
Organizers:
Artificial Intelligence Association of Thailand (AIAT),
Sirindhorn International Institute of Technology (SIIT),
Thammasat University (TU),
Prince of Songkla University (PSU),
National Electronics and Computer Technology Center
(NECTEC)
Printing Production: VS 8 Inter Limited Partnership,
2/484 Moo 1, Klong Luang, Pathum Thani, 12120, Thailand
Date Published:
August 2016
Publisher:
Artificial Intelligence Association of Thailand (AIAT), 22/1, Soi Phutta Bucha 30, Phutta Bucha Rd. Bang mot, Tungkru, Bangkok, 10140, Thailand
Email: [email protected], URL: http://aiat.in.th/
Contact:
Sirindhorn International Institute of Technology (SIIT),
Thammasat University (TU), 131 Moo 5 Tiwanont Road, Bangkadi, Muang Pathumthani 12000, Thailand
Tel. +66-2-501-3505(-20) Fax +66-2-501-3524
Email: [email protected]
© 2016 by AIAT, Artificial Intelligence Association of Thailand (PRICAI 2016)
Printed in Thailand
ISBN 978-616-92700-1-0
Table of Contents
Welcome Message from the PRICAI/PRIMA General Chairs��1
Welcome Message from the PRICAI Program Chairs��5
Welcome Message from the PRIMA Genaral/Program Chairs��7
Welcome Message from the PKAW Workshop Chairs��9
Message from the AI4T Workshop Chairs��10
Message from the AIED Workshop Chairs��11
Message from the I3A Workshop Chairs��12
Message from the IWEC Workshop Chairs��13
Message from the RSAI Workshop Chairs��14
Message for Special Track on Smart Modelling and Simulation Chairs (SMS)��16
PRICAI Organization��17
PRIMA Organization��25
PKAW Organization��30
PRICAI Workshop Organization��32
Special Track on SMS Organization��33
Organizers/ Sponsors/ Supporters��34
PRICAI Sessions��36
PRIMA Sessions��37
PRIMA Tutorial Sessions (Mini-School)��37
PKAW Workshop Sessions��38
PRICAI Workshop Sessions��39
PRICAI Tutorial Sessions��39
Overall Schedule��40
Detailed Schedule��45
Conference Venue��72
Keynote/Invited Speakers��77
IWEC Invited Speaker��85
PRICAI Tutorials��86
PRIMA Tutorials (Mini-School)��96
Special Track (SMS)��104
Abstracts��106
PRICAI Abstracts��107
PRICAI Workshop (AI4T, AIED, I3A, IWEC, RSAI, PeHealth) Abstracts��163
PRIMA Abstracts��200
PRIMA Student Abstracts��221
PKAW Abstracts��226
Proceedings Online Access (Springer)��239
Guides��240
Welcome Message from the PRICAI/PRIMA General
Chairs
The Pacific Rim International Conference on Artificial Intelligence (PRICAI) is a series
of biennial international conferences which concentrate on AI theories, technologies
and their applications in the areas of social and economic importance for countries in
the Pacific Rim. In the past, the conferences have been held in Nagoya (1990), Seoul
(1992), Beijing (1994), Cairns (1996), Singapore (1998), Melbourne (2000), Tokyo
(2002), Auckland (2004), Guilin (2006), Hanoi (2008), Daegu (2010), Kuching (2012),
and Gold Coast (2014). The series aims to strengthen artificial intelligence research
community, which includes researchers, educators, practitioners, and users, by providing a stage exchange ideas and research results, as well as to provide opportunities to
collaborate among several parties in artificial intelligence research. The Fourteenth Pacific Rim International Conference on Artificial Intelligence (PRICAI 2016) will be held
at Hilton Phuket Arcadia Resort & Spa, Phuket, Thailand during August 22-26, 2016.
Together with PRICAI 2016, we have the 19th International Conference on Principles and Practice of Multi-Agent Systems (PRIMA 2016) as a collocated conference. Started as an Asia-Pacific workshop in 1998 and run as a full conference
since 2009, PRIMA has become one of the leading and influential scientific conferences for research on multi-agent systems. Each year, PRIMA brings together researchers, developers, and practitioners from academia and industry to show-case
research in several domains, ranging from foundations of agent theory and engineering aspects of agent systems, to emerging interdisciplinary areas of agent-based
research. Previous successful editions were held in Nagoya, Japan (2009), Kolkata, India (2010), Wollongong, Australia (2011), Kuching, Malaysia (2012), Dunedin, New Zealand (2013), Gold Coast, Australia (2014), and Bertinoro, Italy (2015).
As parts of PRICAI 2016, we have one major workshop, namely the 2016 Pacific Rim
Knowledge Acquisition Workshop (PKAW 2016), as well as six satellite workshops; the
Workshop on eHealth Mining (PeHealth 2016), the Workshop on Image, Information,
and Intelligent Applications (I3A 2016), the Workshop on Artificial Intelligence for Educational Applications (AIED 2016), the Workshop on Artificial Intelligence for Tourism
(AI4T 2016), the 7th International Workshop on Empathic Computing (IWEC 2016) and
the Research Student Symposium on the Artificial Intelligence and Applications (RSAI
2016). In addition to the workshops, there is one special track called Special Track on
Smart Modelling and Simulation, including in the PRICAI sessions. The four PRICAI 2016 tutorials are (1) Deep learning for natural language processing by Hidekazu
Yanagimoto (Osaka Prefecture University, Japan), (2) Introduction to Coalition Formation and its Application by Chattrakul Sombattheera (Mahasarakham University, Thailand), (3) Collective Intelligence: Using Systems/Design Thinking Methods to Improve
Group Intelligence by Jarun Ngamvirojcharoen (Sertis Corporation, USA) and Komes
1
Chandavimol (Data Science Thailand) and (4) Visually See Text Mining Math Processes on LSA, SVD, and Gibbs Sampling by Yukari Shirota (Gakushuin University, Japan).
As parts of PRIMA 2016, a special tutorial, running as a mini-school on
multi-agent systems, is arranged with a number of prominent tutorial instructors; Enrico Pontelli (New Mexico State University, USA), Pradeep Varakantham (Singapore Management University, Singapore), Makoto Yokoo (Kyushu
University, Japan), and Aditya Ghose (University of Wollongong, Australia)
As for the PRICAI/PRIMA keynote speeches, five titles are (1) Global Data Warming
for AI Spring by Sheng-Chuan Wu (Franz Inc., Silicon Valley, USA), (2) From AdaBoost to Optimal Margin Distribution Machines by Zhi-Hua Zhou (Nanjing University,
China), (3) Agent-based modelling and simulation for co-operative traffic and transport
by Joerg P. Mueller (Clausthal University of Technology, Germany), (4) Intercultural
Collaboration: Human-Aware Research on Multiagent Systems by Toru Ishida (Kyoto University, Japan), and (5) Argumentation for Practical Reasoning by Phan Minh
Dung (Asian Institute of Technology, Thailand). The first two speakers are invited
by PRICAI, the next two by PRIMA, and the last one by both PRICAI and PRIMA.
We wish to express our gratitude to the PRICAI Program Co-Chairs; Richard Booth
(Cardiff University, UK) and Min-Ling Zhang (Southeast University, China) and the
PRIMA Program Co-Chairs; Matteo Baldoni (University of Torino, Italy), Amit K.
Chopra (Lancaster University, United Kingdom), Tran Cao Son (New Mexico State University, US), Michael Mäs (University of Groningen, The Netherlands) for their continued support on program-related activities. We also thank to the PRICAI/PRIMA 2016
program committee and reviewers works hard on the collection of submissions and the
review process of the technical papers on substantial, original, and unpublished research
in all aspects of artificial intelligence and multi-agent systems. We are also thankful to
the Tutorial Co-Chairs Sankalp Khanna (CSIRO, Australia), Manabu Okumura (Tokyo
Institute of Technology, Japan) and Kritsada Sriphaew (Rangsit University, Thailand)
for selecting the fruitful tutorials, as well as the mini-school arrangement chairs; William Yeoh (New Mexico State University, USA) and Bo An (Nanyang Technological
University, Singapore) for their kind support on the arrangement of the PRIMA minischool. We appreciated the PRICAI/PRIMA Workshop Co-Chairs, Masayuki Numao
(Osaka University, Japan), Boonserm Kijsirikul (Chulalongkorn University, Thailand),
Sanparith Marukatat (NECTEC, Thailand) and Jamal Bentahar (Concordia University,
Canada) for coordinating the attractive workshops. We also would like to thank the
PKAW honorary chairs; Paul Compton (University of New South Wales, Australia),
Hiroshi Motoda (Osaka University, Japan) and workshop co-chairs; Hayato Ohwada
(Tokyo University of Science, Japan) and Kenichi Yoshida (University of Tsukuba,
Japan). Special thanks to workshop organizers/co-coordinators, particularly Chuleerat
Jaruskulchai (KU, Kasetsart University, Thailand) for peHealth, Narit Hnoohom (Mahidol University, Thailand) for I3A, Thepchai Supnithi (NECTEC, Thailand) and Rachada Kongkrachandra (Thammasat University, Thailand) for AIED, Manabu Okumura
2
(Tokyo Institute of Technology, Japan) for AI4T, Merlin Teodosia Suarez ((De La Salle
University, Philippines) for IWEC, Mahasak Ketcham and Thaweesak Yingthawornsuk
(King Mongkut’s University of Technology North Bangkok, Thailand) for RSAI, and
Patiyuth Pramkeaw (KMUTT, Thailand) for special session and doctoral symposium.
We are indebted the sponsorship co-chairs; Chai Wutiwiwatchai (NECTEC, Thailand)
and Mahasak Ketcham (IT, KMUTNB) as well as our program chair Richard Booth
(Cardiff University, UK) for their kind help in finding a sponsorship fund to our conference. Thanks to Waralak Vongdoiwang Siricharoen (UTCC, Thailand), Vincent C.S.
Lee (Monash University, Australia), Nadin Kokciyan (Bogazici University, Turkey),
Tenda Okimoto (Kobe University, Japan), and Jantima Polpinij (Mahasarakham University, Thailand) for their help in publicity, Neil Yorke-Smith (American University
of Beirut, Lebanon) for the help in publications, Chutima Beokhaimook (Rangsit University, Thailand), Nongnuch Ketui (RMUTL, Nan, Thailand) and Choermath Hongakkaraphan (SIIT, Thammasat University) for their support on financial matters, and
Kobkrit Viriyayudhakorn (SIIT, Thammasat University, Thailand), Thanasan Tanhermhong (SIIT, Thammasat University, Thailand), Wirat Chinnan (SIIT, Thammasat University, Thailand), and Federico Capuzzimati (University of Torino, Italy) for helping
us as the web masters. These conferences cannot be implemented without the support
of our local organizing co-chairs; Rattana Wetprasit (Prince of Songkla University,
Thailand), Virach Sortlertlamvanich (SIIT, TU., Thailand), Thepchai Supnithi (NECTEC, Thailand), and Nattapong Tongtep (Prince of Songkla University, Thailand) and
our secretary generals; Thatsanee Chareonporn (Burapa University, Thailand), Choermath Hongakkaraphan (SIIT, Thammasat University), and Kiyota Hashimoto (Prince
of Songkla University, Thailand). Thanks to PRICAI/PRIMA steering committee.
Last but not the least; we would like to thank the Honorary Co-Chairs; Abdul Sattar (Institute for Integrated and Intelligent Systems, Griffith University, Australia), Hiroshi Motoda
(Osaka University, Japan), Wai Kiang (Albert) Yeap (AUT University, New Zealand), Vilas
Wuwongse (Mahidol University, Thailand), Somnuk Tangtermsirikul (SIIT, Thammasat
University, Thailand), Sarun Sumriddetchkajorn (NECTEC, Thailand), and Pun Thongchunum (Prince of Songkla University, Thailand) for their kind support and guidance.
We gratefully acknowledge the support of the organizing institutions; Artificial Intelligence Association of Thailand, Thammasat University, Prince of Songkla University
and NECTEC, as well as the financial support from Artificial Intelligence Journal (AIJ),
Air Force Office of Scientific Research (AFOSR), Asian Office of Aerospace Research
and Development (AOARD), International Foundation for Autonomous Agents and
Multiagent Systems (IFAAMAS), Thammasat University (TU), Thailand Convention
and Exhibition Bureau (TCEB), SERTIS Co., Ltd., Defence Technology Institute (DTI),
Provincial Electricity Authority of Thailand (PEA), Franz Inc., and Springer Publishing. Special thanks to Easy chair, whose paper submission platform we used to organize
reviews and collate the files for this proceedings. We are also grateful to Springer, its
computer science Vice-President Publishing, Alfred Hofmann, and Anna Kramer, for
3
their assistance in publishing the PRICAI 2016 proceedings and the PRIMA 2016 proceedings, respectively, as a volume in its Lecture Notes in Artificial Intelligence series.
Finally we wish to thank the keynote speakers, PRICAI/PRIMA/PKAW/Workshop authors, tutorial instructors and mini-school organizers who help in the PRICAI/PRIMA2016
for their contribution and support. We hope all participants took this opportunity to share
and exchange ideas with the others and enjoyed PRICAI/PRIMA 2016 in Phuket, Thailand
August 2016
PRICAI 2016 General Co-Chairs
Thanaruk Theeramunkong, SIIT, Thammasat University, Thailand
Dickson Lukose, MIMOS Berhad, Malaysia
PRIMA 2016 General Co-Chairs
Katsutoshi Hirayama, Kobe University, Japan
Paolo Torroni, University of Bologna, Italy
Honorary Chair Representatives
Abdul Sattar, Griffith University, Australia
Hiroshi Motoda, Osaka University, Japan
4
Welcome Message from the PRICAI Program Chairs
This volume contains the papers presented at the 14th Pacific Rim International Conference on Artificial Intelligence (PRICAI 2016) held during August 22-26, 2016 in Phuket,
Thailand. PRICAI is a biennial conference inaugurated in Tokyo in 1990. It provides a
common forum for researchers and practitioners in various branches of artificial intelligence (AI) to exchange new ideas and share experience and expertise. Over the past 26
years the conference has grown, both in participation and scope, to be a premier international AI event for all major Pacific Rim nations as well as countries from further afield.
This year marked the first time that the conference was held in Thailand. In addition to the
main track, PRICAI 2016 featured a special track on Smart Modelling and Simulation.
This year, we received 161 high-quality submissions from 25 countries to both the main
and special tracks. From these, 53 papers (33%) were accepted as regular papers, with a
further 15 accepted as short papers. Each submitted paper was considered by the Program
Committee members and external reviewers, and evaluated against criteria such as relevance, significance, technical soundness, novelty and clarity. Every paper received at least
three reviews, and in some cases up to four, supplemented by rigorous discussion among the
reviewers. Finally, the program co-chairs read the reviews and discussion among reviewers, and made the final decision to ensure fairness and consistency in the paper selection.
The technical program began with two days of workshops and tutorials, followed
by the main conference program. The workshops included the Pacific Rim Knowledge Acquisition Workshop (PKAW), co-chaired by Hayato Ohwada (Tokyo University of Science, Japan) and Kenichi Yoshida (University of Tsukuba, Japan),
which has long enjoyed a successful co-location with PRICAI. Authors of short papers presented their results during poster sessions, but were also given opportunity to give shortened talks to introduce their work. As in previous years, participants
at PRICAI were also able to attend the co-located 19th International Conference on
Principles and Practice of Multi-Agent Systems (PRIMA 2016). We were honored
to have three outstanding keynote speakers, whose contributions have crossed discipline boundaries: Phan Minh Dung (Asian Institute of Technology, Thailand),
Sheng-Chuan Wu (Franz Inc., USA) and Zhi Hua Zhou (Nanjing University, China). We are grateful to them for sharing their insights on their latest research with us.
It would not have been possible to organize the technical program without the considerable help of various people who committed their time and effort towards making
5
PRICAI 2016 a success. We would like to thank the Program Committee members and
external reviewers for their engagements in providing rigorous and timely reviews. It is
because of them that the quality of the papers in this volume is maintained at a high level.
We wish to express our gratitude to the General Co-Chairs Dickson Lukose (MIMOS
Berhad, Malaysia) and Thanaruk Theeramunkong (Thammasat University, Thailand) for their continued support and guidance. We are also thankful to the Tutorial
Co-Chairs Sankalp Khanna (CSIRO, Australia), Manabu Okumura (Tokyo Institute
of Technology, Japan) and Kritsada Sriphaew (Rangsit University, Thailand) for selecting the fruitful tutorials, and the Workshop Co-Chairs Masayuki Numao (Osaka
University, Japan), Boonserm Kijsirikul (Chulalongkorn University, Thailand) and
Sanparith Marukatat (NECTEC, Thailand) for coordinating the attractive workshops.
We gratefully acknowledge the support of the organizing institutions Artificial Intelligence Association of Thailand, Thammasat University, Prince of Songkla University and NECTEC, as well as the financial support from Artificial Intelligence
Journal, Air Force Office of Scientific Research, Asian Office of Aerospace Research and Development, Thammasat University, Thailand Convention and Exhibition Bureau, SERTIS Co., Ltd., Defence Technology Institute, Provincial Electricity Authority of Thailand, Electronic Government Agency, Franz Inc., MIMOS
Berhad, and Springer Publishing. Special thanks to Easy chair, whose paper submission platform we used to organize reviews and collate the files for this proceedings.
We are also grateful to Springer, its computer science Vice-President Publishing,
Alfred Hofmann, and Anna Kramer, for their assistance in publishing the PRICAI
2016 proceedings as a volume in its Lecture Notes in Artificial Intelligence series.
Last but not least, we also want to thank all authors and all conference participants
for their contribution and support. We hope all participants took this opportunity to
share and exchange ideas and thoughts with one another and enjoyed PRICAI 2016.
August 2016
PRICAI Program Chairs
Richard Booth, Cardiff University, UK
Min-Ling Zhang, Southeast University, China
6
Welcome Message from the PRIMA Genaral/Program
Chairs
Welcome to the proceedings of the 19th International Conference on Principles and
Practice of Multi-Agent Systems (PRIMA 2016) held in Phuket, Thailand, during August 22-26, 2016. Started as an Asia-Pacific workshop in 1998 and run as a full conference since 2009, PRIMA has become one of the leading and influential scientific
conferences for research on multi-agent systems. Each year, PRIMA brings together
researchers, developers, and practitioners from academia and industry to showcase
research in several domains, ranging from foundations of agent theory and engineering aspects of agent systems, to emerging interdisciplinary areas of agent-based
research. Previous successful editions were held in Nagoya, Japan (2009), Kolkata, India (2010), Wollongong, Australia (2011), Kuching, Malaysia (2012), Dunedin, New Zealand (2013), Gold Coast, Australia (2014), and Bertinoro, Italy (2015).
The 2016 edition was a special one for a number of reasons. One, to foster a larger Asia Pacific community, it was co-located with the 14th Pacific Rim International Conference on Artificial Intelligence (PRICAI 2016). Two, to foster interdisciplinarity, we ran a social science track, whose accepted papers will
be fast-tracked into the Journal of Artificial Societies and Social Simulation.
Three, to foster student participation, we ran a special student session track. Student authors of accepted papers received free registration for the conference.
We received 50 full paper submissions from 22 countries. Each submission was carefully reviewed by at least three members of the Program Committee (PC) composed
of 107 prominent international researchers. The review period was followed by PC
discussions moderated by Senior Program Committee (SPC) members. The PRIMA SPC has been part of the PRIMA reviewing scheme since 2010, and this year it
included 21 members. At the end of the reviewing process, in addition to the technical reviews, each paper received a summary metareview by an SPC member.
The PC and SPC was truly international, involving researchers from 28 countries.
PRIMA 2016 accepted 17 full papers, giving an acceptance rate of 34%, 16 papers are included in this volume. Moreover, the volume contains three extended abstracts, accepted for the presentation in the social science track, and nine
promising early innovation short papers. Further, we accepted seven submissions for the student session track. In addition to paper presentations sessions,
7
the conference also ran a workshop, a mini-school, and three keynote talks.
We would like to thank all individuals, institutions, and sponsors who supported PRIMA 2016. We thank the authors for submitting high-quality research papers,
confirming PRIMA’s reputation as a leading international conference in multi-agent
systems. We are indebted to our SPC and PC members and additional reviewers for
writing insightful reviews and recommendations for the submissions. We are grateful to members of the PRIMA 2016 organizing committee, who worked behind the
scenes to make PRIMA 2016 successful. These include social science track chair Michael Mäs; workshop chairs Jamal Bentahar and Masayuki Numao; publications chair
Neil Yorke-Smith; publicity chairs Nadin Kökciyan and Tenda Okimoto; mini-school
chairs Bo An and William Yeoh; web chair Federico Capuzzimati; finance chairs
Chutima Beokhaimook, Choermath Hongakkaraphan, and Nongnuch Ketui; local organizing chairs Jantima Polpinij, Virach Sortlertlamvanich, Thepchai Supnithi, Nattapong Tongtep, and Rattana Wetprasit. We thank Enrico Pontelli, Pradeep Varakantham, Makoto Yokoo and Aditya Ghose for doing tutorials in the mini school; Jörg
P. Müller, Phan Minh Dung, and Toru Ishisa for the keynotes. Special thanks to some
individuals who have consistently supported this conference, in particular the senior
advisers of PRIMA 2016, Aditya Ghose, Guido Governatori, and Makoto Yokoo.
We are grateful to Elsevier Artificial Intelligence and the International Foundation for
Autonomous Agents and Multiagent Systems for sponsoring PRIMA 2016. We thank the
Journal of Autonomous Agents and Multi-Agent Systems, ACM Transactions on Autonomous and Adaptive Systems, Fundamenta Informaticae, and the International Journal of
Agent-Oriented Software Engineering for agreeing to fast track selected papers.We also
thank EasyChair for the use of their conference management system. Finally, we thank
Springer for publishing the conference proceedings. We hope you enjoy the proceedings!
August 2016
PRIMA General Chairs
Katsutoshi Hirayama, Kobe University, Japan
Paolo Torroni, University of Bologna, Italy
PRIMA Program Chairs
Matteo Baldoni, University of Torino, Italy
Amit Chopra, Lancaster University, UK
Tran Cao Son, New Mexico State University, US
8
Welcome Message from the PKAW Workshop Chairs
This volume contains the papers presented at PKAW2016: The 14th InternationalWorkshop on Knowledge Management and Acquisition for Intelligent Systems, heldduring
August 22-23, 2016 in Phuket, Thailand, in conjunction with the 14th PacificRim International Conference on Artificial Intelligence (PRICAI 2016).In recent years, unprecedented data, called big data, have become available and knowledge acquisition
and learning from big data are increasing in importance. Varioustypes of knowledge
can be acquired not only from human experts but also from diversedata. Simultaneous acquisition from both data and human experts increases its importance. Multidisciplinary research including knowledge engineering, machine learning, natural language
processing, human-computer interaction, and artificial intelligence is required. We
invited authors to submit papers on all aspects of these area. Another important and
related area is applications. Not only in the engineering field but also in the social science field (e.g., economics, social networks, and sociology), recent progress in knowledge acquisition and data engineering techniques is leading to interesting applications.
We invited submissions that present applications tested and deployed in real-life settings. These papers should address lessons learned from application development and
deployment. As a result, a total of 61 papers were considered. Each paper was reviewed
by at least two reviewers, of which 28 % were accepted as regular papers and 8 % as
short papers. papers were revised according to the reviewers’ comments. Thus, this
volume includes 16 regular papers and five short papers. We hope that these selected
papers and the discussion during the workshop lead to new contributions in this research area. The workshop co-chairs would like to thank all those who contributed to
PKAW 2016, including the PKAW Program Committee and other reviewers for their
support and timely review of papers and the PRICAI Organizing Committee for handling all of the administrative and local matters. Thanks to EasyChair for streamlining
the whole process of producing this volume. Particular thanks to those who submitted papers, presented, and attended the workshop. We hope to see you again in 2018.
August 2016
PKAW Workshop Chairs
Hayato Ohwada, Tokyo University of Science,Japan
Kenichi Yoshida, University of Tsukuba, Japan
9
Message from the AI4T Workshop Chairs
The tourism industry has become one of the fastest growing industriesin the world. Furthermore, the travel and tourism industry has always been open to new technologies.With
the increased interest in the travel and tourism industry in the world, developing new information and communication technologies (ICT) in the travel and tourism industry has
become quite important in these days. Since artificial intelligence technologies can be
considered one of the key technologies in ICT, they should play a key role in the travel
and tourism industry.Today, AI-based developments in the field are at the forefront. In
fact, AI developments and researches have induced much change in this industry. We can
expect this innovation to continue: at both the industrial level and the academic level.
This workshop offers a worldwide and unique forum for attendees from academia,
industry, government, and other organizations to actively exchange, share, and
challenge state-of-the-art researches and industrial case studies on the application of artificial intelligence technologies to travel and tourism. I hope you will enjoy the workshop. Once again, thank you very much for joining our workshop. work
August 2016
AI4T Workshop organizers
Manabu Okumura, Tokyo Institute of Technology, Japan
Hidetsugu Nanba, Hiroshima City University, Japan
Kazutaka Shimada, Kyushu Institute of Technology, Japan
Fumito Masui, Kitami Institute of Technology, Japan
ThanarukTheeramunkong, SIIT, Thammasat University, Thailand
10
Message from the AIED Workshop Chairs
The “Artificial Intelligence for Educational Applications” (AIED) Workshop at the 14th Pacific Rim International Conference on Artificial Intelligence (PRICAI 2016) has organized as a forum for researchers who are
interested in research and development of interactive and adaptive learning environments for learners of all ages, across all domains, to exchange several experiences and results from applications of Artificial Intelligent approaches in Education.
This is the first time the AIED 2016 Workshop has been organized. It received ten
submissions which are passed the full double-blind refereeing process reviewed by at
least three program committee. Three full papers and five short papers are accepted.
We would like to thank all people that assisted to make this Workshop possible. Our grateful
thanks to all the members of Program Committee for timely and excellent reviews. Finally,
we thank to all AIED 2016 participants and we hope this Workshop will give some valuable ideas and inspiration to make practical applications for education in the near future.
August 2016
AIED Workshop organizers
Thepchai Supnithi, NECTEC, Thailand
Rachada Kongkrachandra, Thammasat University, Thailand
11
Message from the I3A Workshop Chairs
The adoption of Artificial Intelligent (AI) technology, and in particular of its mostchallenging components like Image and Information which can constitute the basic
building blocks for a variety of applications within the intelligent world. The combination of the image and emerging information technologies such as image retrieval, computer vision, expert systems, social network analysis, and big data analytics lets us transform everyday information into smart knowledge applications.
International Workshop on Image, Information, and Intelligent Applications (I3A)
is at our 1st edition to collocate with the 14th Pacific Rim International Conference
on Artificial Intelligence (PRICAI 2016), held in Phuket, Thailand, August 22-26,
2016. I3A brings together image, information and intelligent applications from a diverse group of people from industry and academia. As artificial intelligence matures,
it is gradually embedded in the environment and goes beyond conventional information to cover a complex and dynamic environment composed of multiple artifacts.
We received 13 submissions from all over the world. After an intense reviewing process
with at least three reviewers for all papers from different countries. Seven full papers and
two short papers are accepted to present in I3A with 54% acceptance ratio. All accepted
papers have illustrated interesting research projects, results and industrial experiences
that describe significant advances in image, information and intelligent computing. We
hope to provide opportunities for all participants for beneficial discussion and future research resource for you. We look forward to seeing all of you in Phuket for this workshop.
August 2016
I3A Workshop organizers
Narit Hnoohom, Mahidol University, Thailand
Tanasanee Phienthrakul, Mahidol University, Thailand
Mingmanas Sivaraksa, Mahidol University, Thailand
Anuchit Jitpattanakul, King Mongkut’s University of Technology North Bangkok , Thailand
Sakorn Mekruksavanich, University of Phayao, Thailand
Ghita Berrada, Twente University, Netherlands
Rozlina Mohamed, University Technology Malaysia, Malaysia
12
Message from the IWEC Workshop Chairs
The goal of the International Workshop on Empathic Computing (IWEC) has been
to bring together researchers in Asia who wish to solve interesting problems of human-machine interaction that considers social signals and emotions via real-world
modalities of body movement and gestures, facial and audio expressions, and non-vocal utterances, as well as wearable devices to measure physiological signals to measure the effects of such interactions.This year’s workshop focuses on the brain signals
and the emotion. The workshop is honored to have Associate Professor Kenji Tanaka, M.D., Ph.D. from the Department of Neuropsychiatry of Keio University School
of Medicine. This is the first time a brain expert will be delivering a talk at IWEC.
The workshop received a number of papers from all over the world. Each paper
was reviewed by international experts in these fields, and only 50% of the submissions were accepted as full papers. The papers accepted to the workshop range from
processing physiological signals, to the effects of emotion during a student’s learning episode, as well as understanding emotional laughter expressions. The participants will learn a breadth of modalities, how these are processed, and how emotions and its expressions play a large part to make user experiences worthwhile.
The worshiop organizers would like to thank all the advisers and the members of the program committee, including all the authors who submitted papers. In addition, we would like to thank the PRICAI Organizers
for their support to IWEC-16. We look forward to a successful workshop.
August 2016
IWEC Workshop organizers
Merlin Teodosia Suarez, De La Salle University, Philippines
The Duy Bui, Vietnam National University, Hanoi, Vietnam
Ma. Mercedes Rodrigo, Ateneo de Manila University, Philippines
Masayuki Numao, Osaka University, Japan
13
Message from the RSAI Workshop Chairs
This is an exciting time to be a part of an Artificial Intelligence Research Student
Symposium. AI technologies and applications have truly entered our everyday lives,
with AI systems in use throughout society. Against this backdrop of AI’s remarkable
success, the First International Research Student Symposium on the Artificial Intelligence and Application (RAI-2016), to be held in Phuket, Thailand between 22 and
26 August 2016, is the first time the flagship international AI conference has been
held in Kingdom of Thailand, and provide the opportunity for Master Students and
Ph.D. candidates and researchers to share and discuss on research work and idea.
These proceedings collect some of the most exciting research taking place in AI today and offer a window into the future. The theme of the workshop this year is
“Artificial Intelligence and Application.” Being held in Phuket, the Pearl of Andaman Ocean, the RAI workshop will feature invited talks, performances and
a technical track dedicated to the exploration of AI’s growing role in Application and research update, both in enriching and producing AI work and injecting
AI into application to make it an elegant and more accessible scientific discipline.
All accepted papers are included in these proceedings and are invited to present in
workshop sessions and posters. Authors of all papers are invited to give oral presentations, where a distinction is made between long talks and short talks based on the
papers’ quality, clarity, and potential relevance to a wider audience. The Program
Committee did a thorough and highly professional job in reviewing all papers submitted to the conference. Every paper received at least two reviews, which were
provided by members of the program committee (PC). The review process for each
paper was overseen by at least one senior program committee (SPC) member, who
monitored reviews and initiated discussion before and after the author feedback period. Each paper was also managed by the workshop chairs (WC), who engaged
in discussions and gave recommendations on the final decision based on the PC
and SPC input. During the review process, author feedback was taken into account
for the final discussion. When necessary, additional reviews were obtained. Finally,
Workshop Chairs, PCs, SPCs and the Reviewers worked closely in making the final
decisions. The end result is a RAI-2016 program of outstanding quality. We wish to
express our deep appreciation to the Workshop Chairs and Program Committee Members for their outstanding and very professional organization of the review process.
14
We also wish to express our sincere thanks to all reviewers for their valuable dedication. Our deep gratitude extends to the organizers of many other workshop programs
of the PRECAI-2016 conference. We are indebted to their great effort and professionalism. The conference organizing committee provided the much-needed support for the
review process. Finally, we wish to thank all authors of submitted technical papers
for contributing to this great RAI Workshop Symposium. With your high quality work
and devotion, RAI will continue its tradition of excellence and leadership in advancing the Artificial Intelligence Application for the upcoming Symposiums in future..
August 2016
RSAI Workshop organizers
Vincent Shin-Mu Tseng, National Cheng Kung University, Thailand
Mahasak Ketcham, King Mongkut’s University of Technology North Bangkok, Thailand
Thaweesak Yingthawornsuk, King Mongkut’s University of Technology Thonburi, Thailand
15
Message for Special Track on Smart Modelling and
Simulation Chairs (SMS)
Computer based modelling and simulation has become useful tools to facilitate humans to understand systems in different domains, such as physics, astrophysics,
chemistry, biology, economics, engineering and social science. A complex system
is featured with a large number of interacting components, whose aggregate activities are nonlinear and self-organized. Complex systems are hard to be simulated or
modelled by using traditional computational approaches due to the complex relationships of components and distributed features of resources, and dynamic work environments. Meanwhile, smart systems such as multi-agent systems have demonstrated advantages and great potentials in modelling and simulating complex systems.
The special track on Smart Modelling and Simulation (SMS) under the PRICAI’16
will bring together researchers in both artificial intelligence and system modelling/
simulation to discuss research challenges and cutting edge techniques in smart simulation and modelling. We hope to provide opportunities for not only AI researchers, but also domain experts who are interested in the applications of AI techniques
in system modelling and simulation. We look forward to seeing all of you in Phuket.
August 2016
SMS Track organizers
Quan Bai, Auckland University of Technology, New Zealand
Minjie Zhang, University of Wollongong, Australia
Takayuki Ito, Nagoya Institute of Technology, Japan
Fenghui Ren, University of Wollongong, Australia
Katsuhide Fujita, Tokyo University of Agriculture and Technology, Japan
16
PRICAI Organization
Steering Committee
Organizing Committee
Tru Hoang Cao
Ho Chi Minh City University of Technology, Vietnam
Honorary Co-Chairs
Aditya Ghose
University of Wollongong, Australia
(PRIMA Representative)
Byeong-Ho Kang
University of Tasmania, Australia
(PKAW representative)
Dickson Lukose
MIMOS Berhad, Malaysia
Hideyuki Nakashima
Future University Hakodate, Japan
Seong-Bae Park
Kyungpook National University, Korea
Duc Nghia Pham
Abdul Sattar
Griffith University, Australia (Chair)
Toby Walsh
NICTA, Australia
Chengqi Zhang
University of Technology Sydney,
Australia
Zhi-Hua Zhou
Nanjing University, China (Secretary)
Wai Kiang (Albert)
Yeap AUT University, New Zealand
Abdul Sattar
Griffith University, Australia
Hiroshi Motoda
Osaka University, Japan
Vilas Wuwongse
Mahidol University, Thailand
Somnuk Tangtermsirikul
SIIT, Thammasat University, Thailand
Sarun Sumriddetchkajor
NECTEC, Thailand
Pun Thongchunum
Prince of Songkla University, Thailand
General Co-Chairs
Dickson Lukose
MIMOS Berhad
Program Commmittee Co-Chairs
Richard Booth
Cardiff University, UK
Min-Ling Zhang
Southeast University, China
17
Tutorial Co-Chairs
Sankalp Khanna
CSIRO Australian e-Health Research
Centre, Australia
Manabu Okumura
Tokyo Tech, Japan
Kritsada Sriphaew
Rangsit U., Thailand
Workshop Co-Chairs
Masayuki Numao
Boonserm Kijsirikul
Chulalongkorn University, Thailand
Sanparith Marukatat
NECTEC, Thailand
Special Session and Doctoral
Symposium Co-Chairs
Abdul Sattar
Mahasak Ketcham
King Mongkut’s University of Technology
North Bangkok, Thailand
Chuleerat Jaruskulchai
Kasetsart University, Thailand
Rachada Kongkachandra
Thammasat Univesity, Thailand
Pokpong Songmuang
Thammasat University, Thailand
Publicity Co-Chairs
Waralak Vongdoiwangsiricharoen
UTCC, Thailand
Vincent C.S. Lee
Monash University, Australia
Sponsorship Co-Chairs
Chai Wutiwiwatchai
NECTEC, Thailand
Mahasak Ketcham
KMUTNB, Thailand
Financial Co-Chairs
Chutima Beokhaimook
Rangsit University, Thailand
Nongnuch Ketui
RMUTL, Nan, Thailand
Choermath Hongakkaraphan
Local Organizing Co-Chairs
Rattana Wetprasit
Virach Sortlertlamvanich
Thepchai Supnithi
NECTEC, Thailand
Nattapong Tongtep
Patiyuth Pramkeaw
KMUTT, Thailand
Narit Hnoohom
18
Secretary Generals
Quan Bai
Auckland University of Technology,
New Zealand
Thatsanee Chareonporn
Burapa University, Thailand
Ghassan Beydoun
SIIT, Thammasat University , Thailand
Kiyota Hashimoto
Ateet Bhalla
Independent consultant, India
Mehul Bhatt
The University of Bremen, Germany
Webmasters
Kobkrit Viriyayudhakorn
Patrice Boursier
University of La Rochelle, France
Thanasan Tanhermhong
Khalil Bouzekri
Wirat Chinnan
The Duy Bui
Vietnam National University, Vietnam
Marut Buranarach
NECTEC, Thailand
Program Committee
Rafael Cabredo
De La Salle University, Philippines
Mohd Sharifuddin Ahmad
Universiti Tenaga Nasional, Malaysia
Tru Cao
Ho Chi Minh City University of Technology, Vietnam
Eriko Aiba
The University of Electro-Communications, Japan
Songcan Chen
Nanjing University of Aeronautics and
Astronautics, China
Pakinee Aimmanee
Akiko Aizawa
NII, Japan
Wu Chen
Southwest University, China
David Albrecht
Yi-Ping Phoebe Chen
La Trobe University, Australia
Arun Anand Sadanandan
Wai Khuen Cheng
Universiti Tunku Abdul Rahman,
Malaysia
Patricia Anthony
Lincoln University, New Zealand
William K. Cheung
Hong Kong Baptist University, China
Judith Azcarraga
Krisana Chinnasarn
Burapha University, Thailand
19
Seungjin Choi
Pohang University of Science and Technology, Korea
Guido Governatori
NICTA, Australia
Michael Granitzer
University of Passau, Germany
Phatthanaphong Chomphuwiset
Mahasarakham University, Thailand
Fikret Gürgen
Bogazici University, Turkey
Jirapun Daengdej
Assumption University, Thailand
Peter Haddawy
Matthew Dailey
AIT, Thailand
Bing Han
Xidian University, China
Enrique de La Hoz
University of Alcalá, Spain
Choochart Haruechaiyasak
NECTEC, Thailand
Andreas Dengel
German Research Center for Artificial
Intelligence, Germany
Tomomichi Hayakawa
Nagoya Institute of Technology, Japan
Xiangjun Dong
Qilu University of Technology, China
Tessai Hayama
Nagaoka University of Technology, Japan
Shyamala Doraisamy
Universiti Putra Malaysia, Malaysia
Juhua Hu
Simon Fraser University, Canada
Duc Duong
Ho Chi Minh City University of Information Technology, Vietnam
Sheng-Jun Huang
Astronautics, China
Vlad Estivill-Castro
Van Nam Huynh
JAIST, Japan
Christian Freksa
University of Bremen, Germany
Masashi Inoue
Yamagata University, Japan
Katsuhide Fujita
Tokyo University of Agriculture and Technology, Japan
Sanjay Jain
National University of Singapore,
Singapore
Naoki Fukuta
Shizuoka University, Japan
Yuan Jiang
Nanjing University, China
Dragan Gamberger
Ruđer Bošković Institute, Croatia
Geun Sik Jo
Inha University, Korea
Wei Ga
Hideaki Kanai
JAIST, Japan
Xiaoying Gao
Victoria University of Wellington,
New Zealand
Ryo Kanamori
Nagoya University, Japan
20
Kee-Eung Kim
KAIST, Korea
Jiamou Liu
New Zealand
Canasai Kruengkrai
NICT, Japan
Liping Liu
Oregon State University, USA
Alfred Krzywicki
University of New South Wales, Australia
Qing Liu
CSIRO, Australia
Satoshi Kurihara
The University of Electro-Communications, Japan
Yazhou Liu
Nanjing University of Science and Technology, China
Young-Bin Kwon
Chung-Ang University, Korea
Xudong Luo
Sun Yat-Sen University, China
Weng Kin Lai
Tunku Abdul Rahman University College,
Malaysia
Michael Maher
Ho-Pun Lam
Data61, CSIRO, Australia
Iván Marsa-Maestre
University of Alcalá, Spain
Roberto Legaspi
The Institute of Statistical Mathematics,
Japan
Eric Martin
Sanparith Marukatat
NECTEC, Thailand
Chun-Hung Li
Riichiro Mizoguchi
JAIST, Japan
Gang Li
Deakin University, Australia
Li Li
Southwest University, China
Muhammad Marwan Muhammad
Fuad
University of Tromsø, Norway
Ming Li
Ekawit Nantajeewarawat
Wu-Jun Li
Nina Narodytska
Carnegie Mellon University, USA
Yu-Feng Li
M.A.Hakim Newton
Beishui Liao
Zhejiang University, China
Shahrul Azman Noah
Universiti Kebangsaan Malaysia,
Malaysia
Lee Hung Liew
Universiti Teknologi MARA, Malaysia
Masayuki Numao
Tek Yong Lim
Multimedia University, Malaysia
21
Hayato Ohwada
Tokyo University of Science, Japan
Rolf Schwitter
Macquarie University, Australia
Manabu Okumura
Tokyo Institute of Technology, Japan
Nazha Selmaoui-Folcher
PPME, University of New Caledonia,
New Caledonia
Mehmet Orgun
Yi-Dong Shen
Institute of Software, CAS, China
Noriko Otani
Tokyo City University, Japan
Soo-Yong Shin
Asan Medical Center, Korea
Vineet Padmanabhan
University of Hyderabad, India
Yanfeng Shu
CSIRO, Australia
Maurice Pagnucco
Hyeyoung Park
Kyungpook National University, Korea
Tony Smith
University of Waikato,
New Zealand
Duc-Nghia Pham
Chattrakul Sombattheera
Mahasarakham Unversity, Thailand
Jantima Polpinij
Safeeullah Soomro
Indus University, Pakistan
Chao Qian
University of Science and Technology of
China, China
Kritsada Sriphaew
Biplav Srivastava
IBM Research, USA
Joel Quinqueton
LIRMM, France
Markus Stumptner
University of South Australia, Australia
Anca Ralescu
University of Cincinnati, USA
Xing Su
Beijing University of Technology, China
Fenghui Ren
Merlin Suarez
Center for Empathic Human-Computer
Interactions, Philippines
Deborah Richards
Wing-Kin Sung
Singapore
Kazumi Saito
Univesity of Shizuoka, Japan
Chiaki Sakama
Wakayama University, Japan
Boontawee Suntisrivaraporn
SIIT, Thmmasat University, Thailand
Nicolas Schwind
AIST, Japan
Thepchai Supnithi
NECTEC, Thailand
22
David Taniar
Chao Yu
Dalian University of Technology, China
Satoshi Tojo
JAIST, Japan
Yang Yu
Kuniaki Uehara
Kobe University, Japan
Zhiwen Yu
South China University of Technology,
China
Ventzeslav Valev
Institute of Mathematics and Informatics,
Bulgarian Academy of Sciences, Bulgaria
Takaya Yuizono
JAIST, Japan
Yi Zeng
Institute of Automation, Chinese Academy of Sciences, China
Miroslav Velev
Aries Design Automation, USA
Waralak Vongdoiwang Siricharoen
University of the Thai Chamber of Commerce, Thailand
De-Chuan Zhan
Toby Walsh
NICTA and UNSW, Australia
Zhi-Hui Zhan
South China University of Technology,
China
Kewen Wang
Chengqi Zhang
Australia
Qi Wang
Northwestern Polytechnical University,
China
Daoqiang Zhang
Astronautics, China
Wei Wang
Du Zhang
Macau University of Science and Technology, Macau, China
Wayne Wobcke
Guandong Xu
Australia
Junping Zhang
FudanUniversity, China
Shichao Zhang
Guangxi Normal University, China
Ming Xu
Xi’an Jiaotong-Liverpool University,
China
Wen Zhang
Institute of Software, Chinese Academy of
Sciences, China
Roland Yap
Singapore
Yu Zhang
Dayong Ye
Swinburne University of Technology,
Australia
Yanchang Zhao
RDataMining.com
23
Xiaofeng Zhu
Guangxi Normal University, China
Munir, Mohsin
Nguyen, Van Doan
Ou, Wei
Pan, Shirui
Polash, Md. Masbaul Alam
Riveret, Regis
Suh, Suwon
Tajvidi, Masoumeh
Tian, Qing
Tischer, Peter
van de Ven, Jasper
Vu, Huy Quan
Wang, Guixiang
Wang, Hanmo
Wang, Liping
Wang, Yuwei
Wang, Zhe
Wang, Zi-Jia
Wu, Jia
Xu, Feng
Zhang, Heng
Zhang, Lefeng
Zhu, Qi
Zhuang, Zhiqiang
Zu, Chen
Xingquan Zhu
Florida Atlantic University, USA
Fuzhen Zhuang
Institute of Computing Technology, Chinese Academy of Sciences, China
Quan Zou
Tianjin University, China
Dominik Ślęzak
University of Warsaw & Infobright Inc.,
Poland
External Reviewers
Afzal, Muhammad Zeshan
Ahmad, Riaz
Aziz, Tarique
Bizid, Imen
Bukhari, Syed Saqib
Chen, Weiyang
Chen, Xiaohong
Demirović, Emir
Gao, Ping
Gao, Qian
Haryanto, Anasthasia Agnes
Kim, Saehoon
Kong, Jie
Lee, Juho
Lee, Kwanyong
Li, Weihua
Li, Xin
Liu, Mingxia
Liu, Xiaofang
Lv, Guohua
Lye, Guang Xing
Matsubara, Takashi
24
PRIMA Organization
Mini-school Chairs
Bo An
Nanyang Technological University,
Singapore
General Chairs
Katsutoshi Hirayama
Will Yeoh
New Mexico State University, US
Paolo Torroni
University of Bologna, Italy
Web Chair
Program Chairs
Federico Capuzzimati
University of Torino, Italy
Matteo Baldoni
University of Torino, Italy
Financial Co-Chairs
Amit K. Chopra
Lancaster University, UK
Chutima Beokhaimook
Tran Cao Son
New Mexico State University, US
SIIT, TU., Thailand
Michael Mäs
University of Groningen,
The Netherlands.
Nongnuch Ketui
RMUTL, Nan, Thailand
Local Organizing Co-Chairs
Workshop Co-Chairs
Jantima Polpinij
Jamal Bentahar
Concordia University, Canada
Virach Sortlertlamvanich
Masayuki Numao
Publications Chair
Thepchai Supnithi
NECTEC, Thailand
Neil Yorke-Smith
American University of Beirut, Lebanon
Nattapong Tongtep
PSU, Thailand
Publicity Co-Chairs
Rattana Wetprasit
PSU, Thailand
Nadin Kokciyan
Bogazici University, Turkey
Secretary Generals
Tenda Okimoto
Thatsanee Chareonporn
Burapa University, Thailand
Kiyota Hashimoto
PSU, Thailand
25
SIIT, TU., Thailand
Marco Montali
KRDB Research Centre, Free University
of Bozen-Bolzano
Program Committee
Enrico Pontelli
New Mexico State University
Senior Program Committee
Sebastian Sardina
RMIT University
Bo An
Nanyang Technological University
Bastin Tony Roy Savarimuthu
University of Otago
Tina Balke
University of Surrey
Bo Yang
Jilin University
Cristina Baroglio
Dipartimento di Informatica, Università
di Torino
Makoto Yokoo
Kyushu University
Michael Winikoff
University of Otago
Rafael H. Bordini
FACIN-PUCRS
Stephen Cranefield
University of Otago
Jie Zhang
Hoa Khanh Dam
University of Wollongong
Program Committee
Athirai A. Irissappane
Mehdi Dastani
Utrecht University
Stéphane Airiau
LAMSADE - Université Paris-Dauphine
Paul Davidsson
Malmö University
Huib Aldewereld
Delft University of Technology
Yves Demazeau
CNRS - LIG
Natasha Alechina
University of Nottingham
Frank Dignum
Utrecht University
Wagdi Alrawagfeh
Memorial University of Newfoundland
Rino Falcone
Institute of Cognitive Sciences and Technologies-CNR
Leila Amgoud
IRIT - CNRS
Zhi Jin
Peking University
Alexander Artikis
NCSR “Demokritos”
Felipe Meneguzzi
Pontifical Catholic University of Rio
Grande do Sul
Fatma Başak Aydemir
University of Trento
Chiara Bassetti
ISTC-CNR
26
Salem Benferhat
Cril, CNRS UMR8188,
Université d’Artois
Moser Fagundes
University Rey Juan Carlos
Michael Fisher
University of Liverpool
Jamal Bentahar
Concordia University
Nicoletta Fornara
Universita della Svizzera Italiana, Lugano
Olivier Boissier
ENS Mines Saint-Etienne
Elise Bonzon
LIPADE - Universite Paris Descartes
Katsuhide Fujita
Tokyo University of Agriculture and Technology
Nils Bulling
Amineh Ghorbani
Patrice Caire
University of Luxembourg, computer
science dpt.
Guido Governatori
Data61
Cristiano Castelfranchi
Institute of Cognitive Sciences and Technologies
Akin Gunay
The Anh Han
Teesside Univeresity
Qingliang Chen
Department of Computer Science, Jinan
University, Guangzhou 510632, China
James Harland
RMIT University
Massimo Cossentino
National Research Council of Italy
Koen Hindriks
Stefania Costantini
Dipartimento di Ingegneria e Scienze
dell’Informazione e Matematica, Univ.
dell’Aquila
Katsutoshi Hirayama
Kobe University
Xiaowei Huang
University of Oxford
Célia Da Costa Pereira
Université Nice Sophia Anipolis
Fuyuki Ishikawa
National Institute of Informatics
Dave De Jonge
Western Sydney University
Wojtek Jamroga
Polish Academy of Sciences
Nirmit Desai
IBM T J Watson Research Center
Yichuan Jiang
Southeast University
Juergen Dix
Clausthal University of Technology
Anup Kalia
North Carolina State University
Esra Erdem
Sabanci University
Sabrina Kirrane
Vienna University of Economics and
Business - WU Wien
27
Yasuhiko Kitamura
Kwansei Gakuin University
Shigeo Matsubara
Kyoto University
Andrew Koster
Samsung Research Institute
Toshihiro Matsui
Nagoya Institute of Technology
Jérôme Lang
LAMSADE
John-Jules Meyer
Utrecht University
Joao Leite
NOVA LINCS, Universidade Nova de
Lisboa
Roberto Micalizio
Universita’ di Torino
Tsunenori Mine
Kyushu University
Churn-Jung Liau
Academia Sinica, Taipei, Taiwan
Pavlos Moraitis
LIPADE, Paris Descartes University
Yuan Liu
Zeinab Noorian
Ryerson University
Chanjuan Liu
Peking University
Timothy Norman
University of Aberdeen
Fenrong Liu
Tsinghua University, Bejing, China
Brian Logan
University of Nottingham
Andrea Omicini
Alma Mater Studiorum–Università di
Bologna
Emiliano Lorini
IRIT
Nir Oren
Xudong Luo
Sun Yat-sen University
Julian Padget
University of Bath
Marco Lützenberger
Technische Universität Berlin / DAI
Labor
Maurice Pagnucco
The University of New South Wales
Odile Papini
LSIS UMR CNRS 7296
Patrick MacAlpine
University of Texas at Austin
Simon Parsons
King’s College London
Samhar Mahmoud
King’s College London
Fabio Patrizi
Free University of Bozen-Bolzano
Elisa Marengo
Faculty of Computer Science, Free University of Bozen-Bolzano
Duy Hoang Pham
Posts and Telecommunications Institute
of Technology
Viviana Mascardi
DIBRIS (Department of Informatics,
Bioengineering, Robotics and System
Engineering), University of GENOVA, IT
Jeremy Pitt
Imperial College London
28
David Pynadath
Institute for Creative Technologies, University of Southern California
Michaël Thomazo
Inria
Andreea Urzica
University Politehnica of Bucharest
Franco Raimondi
Middlesex University
Wamberto Vasconcelos
Department of Computing Science, University of Aberdeen
Surangika Ranathunga
University of Moratuwa
Alessandro Ricci
University of Bologna
Harko Verhagen
Dept. of Computer and Systems Sciences,
Stockholm University
Juan Antonio Rodriguez Aguilar
IIIA-CSIC
Serena Villata
CNRS - Laboratoire d’Informatique, Signaux et Systèmes de Sophia-Antipolis
Luigi Sauro
University of Naples “Federico II”
Mirko Viroli
Università di Bologna
Bastin Tony Roy Savarimuthu
University of Otago
Kewen Wang
Griffith University
Vadim Savenkov
Vienna University of Economics and
Business (WU)
Brendon J. Woodford
Department of Information Science, University of Otago
Torsten Schaub
University of Potsdam
Nitin Yadav
RMIT University
Claudia Schulz
Imperial College London
William Yeoh
New Mexico State University
Francois Schwarzentruber
École normale supérieure de Rennes
Logan Yliniemi
University of Nevada, Reno
Sandip Sen
University of Tulsa
Neil Yorke-Smith
American University of Beirut
Murat Sensoy
Ozyegin University
Fabio Zambetta
RMIT University, Melbourne, Australia
Carles Sierra
IIIA
Leon van der Torre
University of Luxembourg
Leandro Soriano Marcolino
University of Southern California
Thomas Ågotnes
University of Bergen
Leon Sterling
Swinburne University of technology
Yuqing Tang
Carnegie Mellon University
29
PKAW Organization
Xiongcai Cai
University of New South Wales
Honorary Co-Chairs
Aldo Gangemi
University of Paris 13 / CNR-ISTC
Paul Compton
Udo Hahn
Jena University
Hiroshi Motoda
Osaka University and AFOSR/AOARD,
Japan
Nobuhiro Inuzuka
Toshihiro Kamishima
National Institute of Advanced Industrial
Science and Technology
Workshop Co-chairs
Hayato Ohwada
Tokyo University of Science, Japan
Mihye Kim
Catholic University of Daegu
Kenichi Yoshida
University of Tsukuba, Japan
Yang Sok Kim
University of Tasmania
Advisory Committee
Masahiro Kimura
Ryukoku University
Byeong-Ho Kang
School of Computing and Information
Systems, University of Tasmania,
Australia
Alfred Krzywicki
Setsuya Kurahashi
University of Tsukuba
Deborah Richards
Maria Lee
Shih Chien University
Kyongho Min
Program Committee
Nathalie Aussenac-Gilles
IRIT CNRS
Toshiro Minami
Kyushu Institute of Information Sciences
and Kyushu University Library
Quan Bai
Auckland University of Technology
Luke Mirowski
Ghassan Beydoun
James Montgomery
Ivan Bindoff
Tsuyoshi Murata
Tokyo Institute of Technology
30
Kouzou Ohara
Aoyama Gakuin University
Tomonobu Ozaki
Nihon University
Son Bao Pham
College of Technology, VNU
Alun Preece
Cardiff University
Ulrich Reimer
University of Applied Sciences St. Gallen
Kazumi Saito
Univesity of Shizuoka
Derek Sleeman
Vojtech Svatek
University of Economics, Prague
Takao Terano
Shuxiang Xu
Tetsuya Yoshida
Nara Women’s University
31
PRICAI Workshop Organization
AI4T Committee
Remi Barillec
Aston University, United Kingdom
Manabu Okumura
Rozlina Mohamed
University Technology Malaysia,
Malaysia
Hidetsugu Nanba
Hiroshima City University, Japan
IWEC Committee
Kazutaka Shimada
Kyushu Institute of Technology, Japan
Merlin Teodosia Suarez
Fumito Masui
Kitami Institute of Technology, Japan
The Duy Bui
Vietnam National University Hanoi,
Vietnam
AIED Committee
Ma. Mercedes Rodrigo
Ateneo de Manila University, Philippines
Thepchai Supnithi
NECTEC, Thailand
Masayuki Numao
Rachada Kongkrachandra
PeHealth Committee
Tsukasa Hirashima
Hiroshima University, Japan
Chuleerat Jaruskulchai
I3A Committee
Ornuma Thesprasith
Narit Hnoohom
Rey-Long Liu
Tzu Chi University, Hualien, Taiwan,
R.O.C.
Tanasanee Phienthrakul
RSAI Committee
Mingmanas Sivaraksa
Vincent Shin-Mu Tseng
National Cheng Kung University, Tainan,
Taiwan
Anuchit Jitpattanakul
KMUTNB, Thailand
Mahasak Ketcham
KMUTNB, Thailand
Sakorn Mekruksavanich
University of Phayao, Thailand
Thaweesak Yingthawornsuk
KMUTT, Thailand
Ghita Berrada
Twente University, Netherlands
32
Special Track on SMS Organization
SMS Track Chairs
Quan Bai
New Zealand
Minjie Zhang
Takayuki Ito
Fenghui Ren
Katsuhide Fujita
33
Organizers/ Sponsors/ Supporters
Organizers
• Artificial Intelligence Association of Thailand (AIAT)
• Sirindhorn International Institute of Technology (SIIT)
• Thammasat University (TU)
• Prince of Songkla University (PSU)
• Thailand National Electronics and Computer Technology Center (NECTEC),
National Science and Technology Development Agency (NSTDA)
Sponsors
• The Pacific Rim International Conferences on Artificial Intelligence (PRICAI)
• Artificial Intelligence Journal (AIJ)
• Asian Office of Aerospace Research and Development (AOARD)
• Air Force Office of Scientific Research (AFOSR)
• Thammasat University (TU)
• Sertis Co., Ltd.
• Provincial Electricity Authority of Thailand (PEA)
• Thailand Convention and Exhibition Bureau (TCEB)
• Defence Technology Institute (DTI) (Thailand Public Organization)
• Franz Inc.
• The International Foundation for Autonomous Agents and Multi agent SyStems
(IFAAMAS)
• Springer
Supporters
• Lecture Notes in Computer Science (LNCS)
• Springer
34
Organizers
Sponsors
Supporters
35
PRICAI Sessions
Title
Date
Room
AI Applications
Aug 24,
10:30 - 12:00
Ballroom B
BR
Biometric Recognition
Aug 26 ,
10:30 - 12:00
Arcadia 1
CV
Computer Vision
Aug 25,
15:30 - 17:00
Arcadia 1
DM
Data Mining
Aug 26,
10:30 - 12:00
Similan
IM
Image Processing
10:30 - 12:00,
Aug 25
Arcadia 1
IR
Information Retrieval
Aug 24 ,
15:30 - 17:00
Andaman
KR
Knowledge Representation
Aug 24,
10:30 - 12:00
Andaman
Aug 24,
13:30 - 15:00
Ballroom B
Machine Learning Applications
Aug 24,
15:30 - 17:00
Ballroom B
Natural Language Processing
Aug 25,
13:30 - 15:00
Ballroom B
Neural Networks and Feature Selection
Aug 24,
10:30 - 12:00
Arcadia 1
PR
Pattern Recognition
Aug 25,
13:30 - 15:00
Arcadia 1
RL
Reinforcement Learning
Aug 25,
10:30 - 12:00
Ballroom B
Search
Aug 26,
10:30 - 12:00
Andaman
Social Media
Aug 24,
13:30 - 15:00
Andaman
SMS 1
Special track: Smart Modelling and
Simulation 1 (SMS 1)
Aug 24,
13:30 - 15:00
Arcadia 1
SMS 2
Special track: Smart Modelling and
Simulation 2 (SMS 2)
Aug 24,
15:30 - 17:00
Arcadia 1
Textual Analysis
Aug 25,
15:30 - 17:00
Ballroom B
AI-AP
ML Algo Machine Learning Algorithms
ML App
NLP
NN-FS
Search
SM
TA
36
PRIMA Sessions
Title
Date
PM 1
PRIMA Session 1
PM 2
PRIMA Session 2
PM 3
PRIMA Session 3: Social Science Session
PM 4
PRIMA Session 4
PM 5
PRIMA Session 5
PM 6
PRIMA Session 6
PM-S
PRIMA Student Session
Aug 22,
10:30 - 12:00
Aug 22,
13:30 - 15:00
Aug 22,
15:30 - 17:00
Aug 23,
10:30 - 12:00
Aug 23,
10:30 - 12:00
Aug 23,
15:30 - 17:00
Aug 24,
10:30 - 12:00
Room
Ballroom B
Arcadia 2
PRIMA Tutorial Sessions (Mini-School)
Title
Date
PRIMA Tutorial Session (Mini-School) - I
PM-T1 An Introduction to the Logic Programming
Paradigms with an Eye to Agents Design
Aug 24,
13:30 - 17:00
PRIMA Tutorial Session (Mini-School) - II
PM-T2 Agent technology and business process
management: A new synthesis
Aug 25,
10:00 - 13:00
PRIMA Tutorial Session (Mini-School) - III
PM-T3 Market Design: Designing Social System
by Game Theory
PRIMA Tutorial Session (Mini-School) - IV
PM-T4 Sequential Decision Making for Improving
Efficiency in Urban Environments
Aug 25,
13:30 - 17:00
37
Aug 26,
10:00 - 13:00
Room
Arcadia 2
PKAW Workshop Sessions
Title
Date
PK 1
Knowledge Acquisition from Network and Big
Data - I
Aug 22,
10:30 - 12:00
PK 2
Knowledge Acquisition and Natural Language
Processing
Aug 22,
13:30 - 15:00
PK 3
Knowledge Acquisition and Machine Learning
-I
Aug 22,
15:30 - 17:30
Knowledge Acquisition from Network and Big
PK 4
Data - II
Aug 23,
10:30 - 12:00
PK 5 Knowledge Acquisition and Applications
Aug 23,
13:30 - 15:00
PK 6
Knowledge Acquisition and Machine Learning
- II
38
Aug 23,
15:30 - 17:30
Room
Andaman
PRICAI Workshop Sessions
Title
Date
Room
International Workshop on Image, Information,
WS 1 and Intelligent Applications
(I3A)
Aug 22,
10:30 - 17:00
Arcadia 1
Int’l Research Student Symposium on the ArtiWS 2 ficial Intelligence and Applications
(RSAI+PeHealth)
Aug 22,
10:30 - 17:00
Arcadia 2
International Workshop on Artificial IntelliWS 3 gence for Tourism
(AI4Tourism)
Aug 23,
10:30 - 12:00
Arcadia 1
International Workshop on Artificial IntelliWS 4 gence for Educational Applications
(AIED)
Aug 23,
13:30 - 17:00
Arcadia 1
The 7th International Workshop on Empathic
WS 5 Computing
(IWEC)
Aug 23,
13:30 - 18:00
Arcadia 2
Title
Date
Room
T-1
Deep learning for natural language processing
Aug 22,
10:00 - 17:00
Similan
T-2
Introduction to Coalition Formation and its
Application
Aug 23,
10:00 - 12:30
Arcadia 2
T-3
Aug 23,
Collective Intelligence: Using Systems/Design
Thinking Methods to Improve Group Intelligence 10:00 - 12:30
Similan
T-4
PVisually See Text Mining Math Processes on
LSA, SVD, and Gibbs Sampling
Aug 23,
13:30 - 17:00
Similan
PRICAI Tutorial Sessions
39
Overall Schedule
Monday, August 22, 2016 (Day 1)
08:00 - 09:00
09:00 - 10:00
10:00 - 10:30
Registration
PRIMA/PKAW Opening Ceremony (08:40 - 09:00)
Ballroom B
PRIMA Keynote Speech
Break
PRIMA/PKAW/Workshop
10:30 - 12:00
Ballroom B Andaman Arcadia 1
PM 1
12:00 - 13:30
WS 1
(I3A)
PK 1
Similan
WS 2
(RSAI)
T-1
Lunch
PRIMA/PKAW/Workshop
Ballroom B Andaman Arcadia 1 Arcadia 2
13:30 - 15:00
PM 2
15:00 - 15:30
Arcadia 2
WS 1
(I3A)
PK 2
WS 2
(RSAI)
Break
PRIMA/PKAW/Workshop
15:30 - 17:00
PM 3
WS 1
(I3A)
PK 3
Free Time
18:30 - 21:00
40
WS 2
(RSAI)
Similan
T-1
Similan
T-1
Tuesday, August 23, 2016 (Day 2)
Registration
Ballroom B
PRIMA Keynote Speech
Break
08:00 - 09:00
09:00 - 10:00
10:00 - 10:30
PRIMA/PKAW/Workshop
10:30 - 12:00
Ballroom B Andaman Arcadia 1
PM 4
12:00 - 13:30
13:30 - 15:00
WS 4
(AIED)
PK 5
T-2
T-3
WS 5
(IWEC)
Break
PRIMA/PKAW/Workshop
15:30 - 17:00
PM 6
18:30 - 21:00
Similan
Lunch
PRIMA/PKAW/Workshop
PM 5
15:00 - 15:30
WS 3
(AI4T)
PK 4
Arcadia 2
WS 4
(AIED)
PK 6
WS 5
(IWEC)
(15:30 - 18:00)
Welcome Reception (Lagoon Lawn)
(If rain, Ballroom B) (18:30 - 21:00)
41
Similan
T-4
Similan
T-4
Wednesday, August 24, 2016 (Day 3)
Registration
Opening Ceremony (08:40 - 09:00)
Ballroom B
PRICAI Keynote Speech
Break
08:00 - 09:00
09:00 - 10:00
10:00 - 10:30
PRICAI Session 1
10:30 - 12:00
Ballroom B
Andaman
Arcadia 1
Arcadia 2
AI-AP
KR
NN-FS
PM-S
12:00 - 13:30
13:30 - 15:00
Ballroom B
ML Algo
15:00 - 15:30
15:30 - 17:00
Ballroom B
ML App
18:30 - 21:00
Lunch
PRICAI Session 2
Andaman
Arcadia 1
SM
SMS 1
Break
PRICAI Session 3
Andaman
Arcadia 1
IR
SMS 2
Banquet (Ballroom B)
(18:30 - 21:00)
42
Arcadia 2
PM-T1
(Mini-School)
Arcadia 2
PM-T1
(Mini-School)
Thursday, August 25, 2016 (Day 4)
Registration
Ballroom B
PRICAI Keynote Speech
Break
Ballroom Foyer
08:30 - 09:00
09:00 - 10:00
10:00 - 10:30
PRICAI Session 4
Ballroom B Arcadia 1
10:30 - 12:00
RL
12:00 - 13:30
13:30 - 15:00
15:30 - 17:00
PM-T2
IM
(Mini-School)
(10:30 - 13:00)
Lunch
PRICAI Session 5
Ballroom B Arcadia 1 Arcadia 2
NLP
15:00 - 15:30
Arcadia 2
PM-T3
PR
(Mini-School)
Break
PRICAI Session 6
Ballroom B Arcadia 1 Arcadia 2
TA
PM-T3
CV
(Mini-School)
Free Time
18:30 - 21:00
43
PRICAI Poster
(Place poster:
09:00-10:00)
(Present: 10:00 - 12:30)
Friday, August 25, 2016 (Day 5)
Registration
Arcadia Hall
PRIMA/PRICAI Keynote Speech
Break
08:30 - 09:00
09:00 - 10:00
10:00 - 10:30
PRICAI Session 7
10:30 - 12:00
Ballroom B
Andaman
Arcadia 1
Arcadia 2
Search
DM
BR
(Mini-School)
PM-T4
(10:30 - 13:00)
12:00 - 13:30
Lunch
13:30 - 17:30
Excursion
(Need to register at the registration desk within
August 24, 2016)
44
Detailed Schedule
(Please be notified the page number where the abstract is located
from the number in each blanket at each paper in the schedule)
45
DAY 1
Timetable
MONDAY, AUGUST 22
Ballroom B
Andaman
08:00 - 08:40
Registration
08:40 - 09:00
PRIMA/PKAW Opening Ceremony
09:00 - 10:00
Intercultural Collaboration: Human-Aware Research on Multiagent Systems
(Ballroom B)
Toru Ishida
Department of Social Informatics, Kyoto University
[email protected]
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 1)
PRIMA Session 1
(Session Chair: Aditya Ghose)
PKAW Session 1
(Session Chair: Hayato Ohwada)
Verifying Real-Time Properties
of Multi-Agent Systems via
SMT-based Boun ded Model
Checking (p.220)
Agnieszka M. Zbrzezny and Andrzej Zbrzezny.
Competition Detection from Online
News (p.231)
Zhong-Yong Chen and Chien Chin
Chen.
Competitive VCG Redistribution
Mechanisms for Public Project
Problem (p.207)
Mingyu Guo.
Acquiring Seasonal/Agricultural
Knowledge from Social Media
(p.228)
Hiroshi Uehara and Kenichi Yoshida.
Coalition Structure Formation
Using Anytime Dynamic Programming (p.207)
Narayan Changder, Aditya K.
Ghose, and Animesh Dutta.
Amalgamating Social Media Data
and Movie Recommendation
(p.229)
Maria R. Lee, Tsung Teng Chen, and
Ying Shun Cai.
A Multi Agent System for Understanding the Impact of Technology Transfer Offices in Green-IT
(p.201)
Christina Herzog, Jean-Marc
Pierson and Laurent Lefévre.
Stable Matching in Structured
Networks (p.237)
Ying Ling, Tao Wan, and Zengchang
Qin.
Demand Response Integration
through Agent-based Coordination of Consumers in Virtual
Power Plants (p.208)
Anders Clausen, Aisha Umair,
Zheng Ma, and Bo Nørregaard
Jørgensen.
12:00 - 13:30
LUNCH
46
DAY 1
Timetable
MONDAY, AUGUST 22
Arcadia 1
Arcadia 2
08:00 - 08:40
Similan
Registration
08:40 - 09:00
PRIMA/PKAW Opening Ceremony
09:00 - 10:00
Toru Ishida
[email protected]
(Keynote Speech)
(Ballroom B)
10:00 - 10:30
10:30 - 12:00
(Session 1)
BREAK
Workshop 1 - I3A (1)
(Session Chair: Chuleerat
Jaruskulchai)
Workshop 2 -RSAI (1)
(Session Chair: Pokpong Songmuang)
Learning Latent Word Representations for Enhanced
Short Text Classification
Luepol Pipanmaekaporn and
Suwatchai Kamolsantiroj
(10:30-10:50) (p.187)
Exploring the Distributional
Semantic Relation for ADR and
Therapeutic Indication Identification in EMR (p.182)
Siriwon Taewijit, Thanaruk Theeramunkong (10:30 - 10:50)
A Regression-based SVD
Parallelization using Overlapping Folds for Textual
Data (p.166)
Uraiwan Buatoom, Thanaruk
Theeramunkong and Waree
Kongprawechnon
(10:50-11:10)
Verifying Properties of Multiagent Systems via Bounded Model
Checking (p.198)
Agnieszka Zbrzezny (11:50-11:00)
Desktop Tower Defense is
NP-Hard (p.176)
Vasin Suttichaya
(11:10-11:30)
Enhancement of Palm-Leaf Manuscript for Segmentation
Siriya Phattarachairawee, Montean
Rattanasiriwongwut, Mahasak Ketcham (11:10 - 11:20) (p.179)
Variant Annotation and Clinical interpretation software
for Cancer (VARCIN): Report generating software for
targeted therapy method
Mingmanas Sivaraksa, Pitinat
Asawasutsakorn, Benjamard
Meeboon, Natini Jinawath
(11:30-11:50) (p.197)
MOEPSO for Multi-objective Optimization (p.190)
Ittikon Thammachantuek, Mahasak
Ketcham (11:00 - 11:10)
Comparison of Edge Detection Algorithms for Coastline
Detection in Satellite Imageries
(p.173)
Chutiwan Boonarchatong, Sucha
Smanchat, Mahasak Ketcham,
Nawaporn Wisitpongphan
Fuzzy C-Means Classification (11:20 - 11:30)
of Electroencephalography
(EEG) Waves for Robotic
System Time Events and
Control (p. 238)
Ebrahim A. Mattar and Hessa
J. Al-Junaid (11:50-12:10)
12:00 - 13:30
LUNCH
47
Tutorial 1
(1)
T-1/1
Deep
learning
for natural
language
processing
Hidekazu
Yanagimoto
DAY 1
Timetable
13:30 - 15:00
(Session 2)
MONDAY, AUGUST 22
Ballroom B
Andaman
PRIMA Session 2
(Session Chair: Guido Governatori)
PKAW Session 2
(Session Chair: Deborah Richards)
Individually Rational Strategy-proof Social Choice with Exogenous Indifference Sets (p.213)
Mingyu Guo, Yuko Sakurai, Taiki
Todo, and Makoto Yokoo.
Enhanced Rules Application
Order to Stem Affixation, Reduplication and Compounding Words
in Malay Texts (p.232)
Mohamad Nizam Kassim, Mohd
Aizaini Maarof, Anazida Zainal, and
Amirudin Abdul Wahab
A Collaborative Framework for
3D Mapping using Unmanned
Aerial Vehicles (p.201)
Patrick Doherty, Jonas Kvarnström,
Piotr Rudol,Marius Wzorek, Gianpaolo Conte, Cyrille Berger,
Timo Hinzmann, and Thomas
Stastny.
Building a Process Description
Repository with Knowledge Acquisition (p.230)
Diyin Zhou, Hye-Young Paik, Seung
Hwan Ryu, John Shepherd, and Paul
Compton.
Revenue Maximizing Markets for
Zero-Day Exploits (p.217)
Mingyu Guo, Hideaki Hata, and Ali
Babar.
Specialized Review Selection Using
Topic Models (p.236)
Anh Duc Nguyen, Nan Tian, Yue Xu,
and Yuefeng Li.
Generalising Social Structure
using Interval Type-2 Fuzzy Sets
(p.210)
Christopher K. Frantz, Bastin Tony
Roy Savarimuthu, Martin K. Purvis,
and Mariusz Nowostawski.
Quality of Thai to English Machine Translation (p.235)
Seamus Lyons.
BREAK
15:00 - 15:30
48
DAY 1
MONDAY, AUGUST 22
Timetable
Arcadia 1
Arcadia 2
Similan
13:30 - 15:00
Workshop 1 - I3A (2)
(Session Chair: Sakorn
Mekruksavanich)
Workshop 2 - RSAI (2)
(Session Chair: Thaweesak
Yingthawornsuk)
Tutorial 1
(2)
Virtual Reality System
with Smartphone
Application for Height
Exposure (p.199)
Suppanut Nateeraitaiwa,
Narit Hnoohom
(13:30-13:50)
Real-time Snoring Sound
Detecting U Shape Pillow
System using Data Analysis
Algorithm (p.192)
Patiyuth Pramkeaw, Penpichaya Lertritchai, Nipaporn Klangsakulpoontawee
(13:30 - 13:50)
Evolving Public Opinion
Mining Methods on
Decision Support System
in Thai EGovernment
(p.181)
Jeerana Noymanee, Wimol
San-Um, Thanaruk Theeramunkong
(13:50-14:10)
A multi-objective adaptive
Invasive Weed Optimization intelligence approach
for solving DNA sequence
design (p.165)
Qiang Zhang, Gaijing Yang,
Changjun Zhou, Bin Wang
(13:50 - 14:10)
Classification of Diabetic
Retinopathy Stages using
Image Segmentation and
an Artificial Neural Network (p.172)
Narit Hnoohom, Ratikanlaya Tanthuwapathom
(14:10-14:30)
Fatigue Classification of
Military Mission by EEG
signals via Artificial Neural
Network (ANN) (p.184)
Worawut Yimyam, Mahasak
Ketcham
(14:10 - 14:30)
Medicine Recognition
using Intrinsic Geometric
Property from Pill Image
(p.188)
M Ashraful Amin, Md. Zakir Hossan, Tanjina Piash
Proma
(14:30-14:50)
Arrival Time Prediction
and Train Tracking Analysis (p.169)
Somkiat Kosolsombat, Wasit
Limprasert
(14:30 - 15:00)
(Session 2)
BREAK
15:00 - 15:30
49
T-1/2
Deep learning for natural language
processing
Hidekazu
Yanagimoto
DAY 1
MONDAY, AUGUST 22
Timetable
Ballroom B
Andaman
15:30 - 17:00
PRIMA Session 3
(Session Chair: Michael Mäs)
PKAW Session 3
(Session Chair: Byeong Ho Kang)
How, When and Where Can Spatial Segregation Induce Opinion
Polarization? Two Competing
Models (p.212)
Thomas Feliciani, Andreas Flache,
and Jochem Tolsma.
Abbreviation Identification in Clinical Notes with Level-wise Feature
Engineering and Supervised Learning (p.228)
Thi Ngoc Chau Vo, Tru Hoang Cao,
and Tu Bao Ho.
Can Noise in Behavioral Models
Improve Macro-Precisions? An
Empirical Test (p.206)
Michael Mäs and Dirk Helbing.
A New Hybrid Rough Set and Soft
Set Parameter Reduction Method
for Spam E-Mail Classification Task
(p.227)
Masurah Mohamad and Ali Selamat.
Ali Baba and the Thief, Convention Emergence in
Games (p.202)
Xin Sun and Livio Robaldo.
Combining Feature Selection with
Decision Tree Criteria and Neural
Network for Corporate Value Classification (p.231)
Ratna Hidayati, Katsutoshi Kanamori, Ling Feng, and Hayato Ohwada.
(Session 3)
Heuristics on the Data-collecting
Robot Problem with Immediate
Rewards (p.211)
Zhi Xing and Jae C. Oh.
Towards Better Crisis Management in Support Services Organizations Using Fine Grained
Agent Based Simulation (p.220)
Vivek Balaraman, Harshal Hayatnagarkar, Meghendra Singh, and
Mayuri Duggirala.
50
DAY 1
Timetable
15:30 - 17:00
(Session 3)
Arcadia 1
-
MONDAY, AUGUST 22
Arcadia 2
Similan
Workshop 2 - RSAI (3)
(Session Chair: Thaweesak
Yingthawornsuk)
Tutorial 1 (3)
The Limb Leads ECG Signal Analysis in Myocardial
Infarction Patients (p.193)
Anchana Muankid, Mahasak
Ketcham
(15:30 - 15:40)
-
Estimating PSD Characteristics of ECG in Comparison between Normal and
Supraventricular Subjects
(p.180)
Thaweesak Yingthawornsuk,
Siriphan Phetnuam
(15:40 - 16:00)
T-1/3
Deep learning
for natural language processing
Hidekazu Yanagimoto
51
DAY 2
Timetable
TUESDAY, AUGUST 23
Ballroom B
Andaman
08:30 - 09:00
Registration
09:00 - 10:00
Agent-based modelling and simulation for co-operative traffic and
transport
(Keynote Speech)
(Ballroom B)
Jörg P. Müller
Department of Informatics, Clausthal University of Technology
[email protected]
BREAK
10:00 - 10:30
10:30 - 12:00
(Session 4)
PRIMA Session 4
(Session Chair: Paolo Torron)
PKAW Session 4
(Session Chair: Maria Lee)
Argumentation-Based Semantics for Logic Programs with
First-Order Formulae (p.204)
Phan Minh Dung, Tran Cao Son,
and Phan Minh
Thang.
Predicting the Scale of Trending
Topic Diffusion Among Online
Communities (p.235)
Dohyeong Kim, Soyeon Caren Han,
Sungyoung
Lee, and Byeong Ho Kang.
Resistance to Corruption of General Strategic Argumentation
(p.216)
Michael J. Maher.
Finding Reliable Source for Event
Detection Using Evolutionary
Method (p.232)
Raushan Ara Dilruba and Mahmuda
Naznin.
Balancing Rationality and Utility Workflow Interpretation via Social
Networks (p.237)
in Logic-Based Argumentation
with Classical Logic Sentences
Eui Dong Kim and Peter Busch.
and Belief Contraction (p.205)
Ryuta Arisaka and Ken Satoh.
Argumentation Versus Optimization for Supervised Acceptability Learning (p.203)
Hiroyuki Kido.
Improving Motivation in Survey
Participation by Question Reordering (p.233)
Rohit Kumar Singh, Vorapong Suppakitpaisarn, and Ake Osothongs.
Dialectical Proof Procedures for
Probabilistic Abstract Argumentation (p.208)
Phan Minh Thang.
12:00 - 13:30
LUNCH
52
DAY 2
Timetable
Arcadia 1
TUESDAY, AUGUST 23
Arcadia 2
Similan
08:30 - 09:00
Registration
09:00 - 10:00
Agent-based modelling and simulation for co-operative traffic and
transport
(Keynote Speech)
Jörg P. Müller
[email protected]
BREAK
10:00 - 10:30
10:30 - 12:00
(Session 4)
Workshop 3 - AI4T
(Session Chiar: Manabu Okumura)
Tutorial 2
Tutorial 3
Opening Ceremonies (10:30)
Inferring Tourist Behavior
and Purposes of a Twitter
User (p.185)
Yuya Nozawa, Masaki Endo,
Yo Ehara, Masaharu Hirota,
Syohei Yokoyama, Hiroshi
Ishikawa (10:35-11:00)
Two stage travel salesman
model of world tourism
(p.196)
Surafel Luleseged Tilahun, Jean
Medard T Ngnotchouye
(11:00-11:25)
Extracting and Characterizing Functional Communities
in Spatial Networks (p.183)
Takayasu Fushimi, Kazumi
Saito, Tetsuya Ikeda, Kazuhiro
Kazama (11:25-11:50)
T-3
T-2
Introduction to
Coalition Formation and its
Application
Travellers’ Behaviour Analysis Based on Automatically
Identified Attributes from
Travel Blog Entries (p.194)
Kazuki Fujii, Hidetsugu Nanba,
Toshiyuki Takezawa, Aya Ishino, Manabu Okumura, Youhei
Kurata (11:50-12:15)
Closing (12:15)
12:00 - 13:30
LUNCH
53
Collective Intelligence: Using
Systems/Design
Thinking Methods to Improve
Group Intelligence
Jarun
Ngamvirojcharoen,
Komes Chandavimol
DAY 2
TUESDAY, AUGUST 23
Timetable
Ballroom B
Andaman
13:30 - 15:00
PRIMA Session 5
(Session Chair: Matteo Baldoni)
PKAW Session 5
(Session Chair: Takahira Yamaguchi)
(Session 5)
Semantic Reasoning with Uncertain Information from Unreliable Sources (p.218)
Murat Şensoy, Lance Kaplan, and
Geeth de Mel.
Knowledge Acquisition for Learning Analytics: Comparing Teacher-Derived, Algorithm-Derived,
and Hybrid Models in the Moodle
Engagement Analytics Plugin
(p.234)
Danny Y.T. Liu, Deborah Richards,
Phillip Dawson, Jean-Christophe
Froissard, and Amara Atif.
Sequence Semantics for Normative Agents (p.218)
Guido Governatori, Francesco
Olivieri, Erica Calardo, Antonino
Rotolo, and Matteo Cristani.
Building a Mental Health Knowledge Model to Facilitate Decision
Support (p.229)
Bo Hu and Boris Villazon Terrazas.
Distant Group Responsibility in
Multi-Agent Systems (p.209)
Vahid Yazdanpanah and Mehdi
Dastani.
Building a Working Alliance with a
Knowledge Based System Through
an Embodied Conversational Agent
(p.230)
Deborah Richards and Patrina
Caldwell.
Plan Failure Analysis: Formalization and Application in Interactive Planning Through Natural
Language Communication
(p.215)
Chitta Baral, Tran Cao Son, Michael Gelfond, and Arindam Mitra.
BREAK
15:00 - 15:30
54
DAY 2
Timetable
13:30 - 15:00
(Session 5)
Arcadia 1
TUESDAY, AUGUST 23
Workshop 4 - AIED (1)
(Session Chair: Thepchi
Supnithi)
Contents Organization
Support for Logical Presentation Flow (p.175)
Tomoko Kojiri, Yuta Watanabe
(13:30 - 14:00)
K-Mean Algorithm for
Finding Students’ Proficiency with a Framework’s item Examination
(p.186)
Nongnuch Ketui, Kanitha
Homjun, Prasert Luegkhong
(14:00 - 14:20)
Arcadia 2
Similan
Workshop 5 - IWEC (1)
(Session Chair: Masayuki Numao, [Invited Talk],
Juan Lorenzo Hagad,
[Paper Presentation])
Tutorial 4 (1)
Opening Ceremonies
(13:30 - 13:45)
Invited Talk
(13:45 - 14:30)
Recent advances in
emotional and motivational circuits
(p.85)
Kenji Tanaka
Department of Neuropsychiatry Keio
University School of
Medicine
Application of Annotation on Smoothing for
Subject-independent
Examination Timetabling Emotion Recognition
using Prey Predator Algo- based on Electroencephalogram (p.168)
rithm (p.178)
Surafel Luleseged Tilahun, Nattapong Thammasan,
Ken-ichi Fukui, MasayuJean Medard T
ki Numao
Ngnotchouye
(14:30 - 15:00)
(14:20 - 14:40)
Development of Salary
Prediction System to
Improve Student Motivation using Data Mining
Technique (p.177)
Pornthep Khongchai, Pokpong Songmuang
(14:40 - 15:00)
Modeling Negative Affect Detector of Novice
Programming Students using Keyboard
Dynamics and Mouse
Behavior (p.189)
Larry Vea, Ma. Mercedes
Rodrigo
(15:00 - 15:30)
BREAK
15:00 - 15:30
55
T-4/1
Visually See Text
Mining Math
Processes on
LSA, SVD, and
Gibbs Sampling
Yukari
Shirota
DAY 2
TUESDAY, AUGUST 23
Timetable
Room A
15:30 - 17:00
PRIMA Session 6
(Session Chair: Mingyu Guo)
(Session 6)
Room B
PKAW Session 6
(Session Chair: Kenichi Yoshida)
Offer Evaluation and Trade-off
Making in Automated Negotiation Based on Intuitionistic
Fuzzy Constraints (p.215)
Jieyu Zhan and Xudong Luo.
Learning Under Data Shift for
Domain Adaptation: A Model-Based
Co-clustering Transfer Learning
Solution (p.234)
Santosh Kumar, Xiaoying Gao, and
Ian Welch.
Spread of Cooperation in
Complex Agent Networks Based
on Expectation of Cooperation
(p.219)
Ryosuke Shibusawa, Tomoaki Otsuka, and Toshiharu Sugawara
Robust Modified ABC Variant
(JA-ABC5b) for Solving Economic
Environmental Dispatch (EED)
(p.236)
Noorazliza Sulaiman, Junita Mohamad-Saleh, and Abdul Ghani Abro.
Analyzing Topics and Trends in
the PRIMA Literature (p.202)
Hoa Khanh Dam and Aditya K.
Ghose.
Integrating Symbols and Signals
Based on Stream Reasoning and
ROS (p.233)
Takeshi Morita, Yu Sugawara, Ryota
Nishimura, and Takahira Yamaguchi.
Automatic Evacuation Management Using a Multi Agent System and Parallel Meta-heuristic
Search. (p.205)
Leonel Enrique Aguilar Melgar,
Maddegedara Lalith, Tsuyoshi
Ichimura, and Muneo Hori.
Modeling Organizational and Institutional Aspects in Renewable
and Natural Resources Management Context (p.214)
Islem Hènane, Sameh Hadouaj,
Khaled Ghédira, and Ali Ferchichi.
18:30 - 21:00
56
DAY 2
TUESDAY, AUGUST 23
Timetable
Arcadia 1
Arcadia 2
Similan
15:30 - 17:00
Workshop 4 - AIED (2)
(Session Chiar: Ratchada
Kongkachandra)
Workshop 5 - IWEC (2)
(Session Chair: Nattapong
Thammasan)
Tutorial 4 (2)
A Framework to Generate Carrier Path Using
Semantic Similarity of
Competencies in Job
Position (p.164)
Wasan Na Chai, Taneth
Ruangrajitpakorn, Marut
Buranarach, Thepchai
Supnithi
(15:30-16:00)
IWEC Poster
(16:00 - 16:30)
(Session 6)
Automatic Question
Generation on SQL
Language Using Template-Based Method
Jittima Janphat, Orawan
Chaowalit (16:00-16:20)
(p.170)
Affective Laughter
Expressions from Body
Movements (p.167)
Jocelynn Cu, Ma. Beatrice
Luz, McAnjelo Nocum,
Timothy jasper Purganan,
Wing San Wong
(16:30 - 17:00)
T-4/2
Multimodal Latent
Feature Learning for
Psyco-Physiological Stress
Modeling and Detection
(p.191)
Juan Lorenzo Hagad,
Ken-ichi Fukui, Masayuki
Numao (17:00 - 17:20)
TSCS Monitor: Generation of Time Series
Cross Section Tables
from Moodle Logs for
Tracking In-Class Page
Views Using Excel MacComputational Model for
ros (p.195)
Affect Detection in LearnKonomu DOBASHI
ing (p.174)
(16:20-16:40)
Najlaa Mokhtar, Syaheerah
Building a Semantic
Lebai Lutfi
Ontology for Virtu(17:20 - 17:40)
al Peers in Narrative-Based Environments (p.171)
Summary and Closing
Ethel Chua Joy Ong,
(17:40-18:00)
Danielle Grace Consignado, Sabrina Jane Ong,
Zhayne Chong Soriano
(16:40-17:00)
18:30 - 21:00
Visually See
Text Mining
Math Processes
on LSA, SVD,
and Gibbs
Sampling
57
Yukari
Shirota
DAY 3
Timetable
08:00 - 08:40
08:40 - 09:00
09:00 - 10:00
WEDNESDAY, AUGUST 24
Ballroom B
Registration
Opening Ceremony
Andaman
(Ballroom B)
Global Data Warming for AI Spring
Sheng-Chuan Wu
Franz Inc., Silicon Valley, USA
[email protected]
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 1)
AI Applications
(Session Chair: Chattrakul Sombattheera)
Knowledge Representation
(Session Chair: Patrick Doherty)
Setting An Effective Pricing Policy
for Double Auction Marketplaces
(p.156)
Bing Shi, Yalong Huang, Shengwu
Xiong and Enrico H. Gerding.
Computing Probabilistic Assumption-based Argumentation
(p.127)
Nguyen Duy Hung.
Offline Text and Non-text Segmentation for Hand-Drawn Diagrams
(p.150)
Buntita Pravalpruk and Matthew
Dailey.
Restricted Four-valued Semantics for Answer Set Programming
(p.153)
Chen Chen and Zuoquan Lin.
A Study of Players’ Experiences
during Brain Games Play
(p.116)
Faizan Ahmad, Yiqiang Chen,
Shuangquan Wang, Zhenyu Chen,
Jianfei Shen, Lisha Hu and Jindong
Wang.
On Partial Features in the DLF
Family of Description Logics
(p.150)
David Toman and Grant Weddell.
(SHORT) ALLIANCE-ROS: A
Software Architecture on ROS
for Fault-Tolerant Cooperative
Multi-Robot Systems (p.118)
Minglong Li, Zhongxuan Cai, Xiaodong Yi, Zhiyuan Wang, Yanzhen
Wang, Yongjun Zhang and Xuejun
Yang.
12:00 - 13:30
LUNCH
58
DAY 3
Timetable
08:00 - 08:40
08:40 - 09:00
09:00 - 10:00
Arcadia 1
Arcadia 2
Registration
Opening Ceremony
(Ballroom B)
Sheng-Chuan Wu
Franz Inc., Silicon Valley, USA
[email protected]
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 1)
Neural Networks and Feature
(Session Chair: Daoqiang Zhang)
PRIMA-Student
(Session Chair: Tran Cao Son)
A Differential Evolution Approach to Feature Selection and
Instance Selection (p.110)
Jiaheng Wang, Bing Xue, Xiaoying
Gao and Mengjie Zhang.
Selected Methods of Model Checking using SAT and SMT-solvers
Agnieszka Zbrzezny. (p.225)
On the Gradient-Based Sequential Tuning of the Echo State
Network Reservoir Parameters
(p.151)
Sumeth Yuenyong.
Using Canonical Correlation
Analysis For Parallelized Attribute Reduction (p.162)
Lin Shang, Mengting Xu and Ping
Li.
(SHORT) Distributed B-SDLM:
Accelerating the Training
Convergence of Deep Neural
Networks through Parallelism
(p.129)
Shan Sung Liew, Mohamed
Khalil-Hani and Rabia Bakhteri.
12:00 - 13:30
Role of Facial Emotion in Social
Correlation (p.224)
Pankaj Mishra, Rafik Hadfi and
Takayuki Ito.
Real-time Collision Handling in
Railway Network: An Agent-based
Approach (p.223)
Poulami Dalapati and Animesh Dutta.
Multi-Agent Based Scalable and
Context Aware Middleware for Typical IoT Scenarios (p.223)
Bikash Choudhury, Subhrabrata
Choudhury and Animesh Dutta.
Security and Access Control in
Multi-AgentSatellite Systems
(p.224)
Pratik Sinha and Animesh Dutta.
Freight Train Scheduling Problem
(p.222)
Samriddhi Sarkar and Animesh Dutta.
A Kernelization approach for Anytime Coalition Structure Generation using Knuth X algorithm
Narayan Changder, Animesh Dutta
and Aditya Ghose. (p.222)
LUNCH
59
DAY 3
Timetable
Ballroom B
Andaman
13:30 - 15:00
Machine Learning Algorithms
(Session Chair: Kazumi Saito)
Social Media
(Session Chair: Quan Zou)
A Novel Isolation-Based Outlier
Detection Method (p.113)
Yanhui Shen, Huawen Liu, Yanxia
Wang, Zhonglong Zheng, Zhongyu
Chen and Guanghua Sun.
An Analysis of Influential Users for
Predicting the Popularity of News
Tweets (p.119)
Krissada Maleewong.
Maximum Margin Tree Error
Correcting Output Codes (p.145)
Fa Zheng, Hui Xue, Xiaohong
Chen and Yunyun Wang.
Modeling of Travel Behavior Processes from Social Media (p.146)
Yuki Yamagishi, Kazumi Saito and
Tetsuo Ikeda.
Multi-view Representative and
Informative Induced Active
Learning (p.149)
Huaxi Huang, Changqing Zhang,
Qinghua Hu and Pengfei Zhu.
A Microblog Hot Topic Detection
Algorithm based on Discrete Particle Swarm Optimization (p.111)
Huifang Ma, Yugang Ji, Xiaohong Li
and Runan Zhou.
(SHORT) Instance Selection
Method for Improving
Graph-based Semi-Supervised
Learning (p.139)
Hai Wang, Shao-Bo Wang and
Yu-Feng Li.
(SHORT) Topic Detection in Group
Chat Based on Implicit Reply
(p.161)
Xinyu Zhang, Ning Zheng, Jian Xu
and Ming Xu.
(Session 2)
15:00 - 15:30
BREAK
60
DAY 3
Timetable
Arcadia 1
Arcadia 2
13:30 - 15:00
Special session SMS 1
(Session Chair: Fenghui Ren)
PRIMA Tutorial 1/1
(Session 2)
Stigmergy-based Influence Maximization in Social Networks
(p.157)
Weihua Li, Quan Bai, Chang Jiang
and Minjie Zhang.
A Concurrent Multiple Negotiation Mechanism under Consideration of A Dynamic Negotiation
Environment (p.109)
Lei Niu, Fenghui Ren and Minjie
Zhang.
Evaluation of Deposit-based
Road Pricing Scheme by Agentbased Simulator (p.131)
Ryo Kanamori, Toshiyuki Yamamoto and Takayuki Morikawa.
PM-T 1/1
PRIMA-Tutorial 1/1
Mini-School
An Introduction to the Logic Programming Paradigms with an Eye
to Agents Design
(SHORT) Capability-aware
Trust Evaluation Model in
Multi-agent Systems (p.125)
Doan Tung Nguyen, Quan Bai and
Weihua Li.
15:00 - 15:30
BREAK
61
Enrico Pontelli
DAY 3
Timetable
Ballroom B
Andaman
15:30 - 17:00
Machine Learning Applications
(Session Chair: Xiaoying Gao)
Information Retrieval
(Session Chair: Yu-Feng Li)
(Session 3)
BDSCyto: an Automated Approach for Identifying Cytokines Based on Best Dimension
Searching (p.124)
Quan Zou, Shixiang Wan, Bing
Han and Zhihui Zhan.
Information Retrieval from Unstructured Arabic Legal Data
(p.138)
Imen Bouaziz Mezghanni and Faiez
Gargouri.
Combining RDR-based Machine
Learning Approach and Human
Expert Knowledge for Phishing
Prediction (p.126)
Hyunsuk Chung, Renjie Chen,
Soyeon Han and Byeong-Ho Kang.
Prediction with Confidence in Item
Based Collaborative Filtering
(p.152)
Tadiparthi V R Himabindu, Vineet
Padmanabhan, Arun K Pujari and
Abdul Sattar.
Learning of Evaluation Functions to Realize Playing Styles in
Shogi (p.142)
Shotaro Omori and Tomoyuki
Kaneko.
Threshold-based Direct Computation of Skyline Objects for Database
with Uncertain Preferences (p.160)
Venkateswara Rao Kagita, Arun K
Pujari, Vineet Padmanabhan, Vikas
Kumar and Sandeep Kumar Sahu.
(SHORT) A Relaxed K-SVD
Algorithm for Spontaneous
Micro-Expression Recognition.
(p.115)
Hao Zheng, Xin Geng and
Zhongxue Yang.
18:30 - 21:00
Banquet (Ballroom B) (18:30 - 21:00)
62
DAY 3
Timetable
Arcadia 1
Arcadia 2
15:30 - 17:00
Special session SMS 2
(Session Chair: Ryo Kanamori)
PRIMA Tutorial 1/2
(Session 3)
L1-Regularized Continuous
Conditional Random Fields
(p.140)
Xishun Wang, Fenghui Ren, Chen
Liu and Minjie Zhang.
Optimization of Road Distribution for Traffic System Based on
Vehicle’s Priority (p.151)
Wen Gu and Takayuki Ito.
Adaptive Learning for Efficient
Emergence of Social Norms in
Networked Multiagent Systems
(p.117)
Chao Yu, Hongtao Lv, Sandip Sen,
Fenghui Ren and Guozhen Tan.
PM-T 1/2
PRIMA-Tutorial 1/2
Mini-School
An Introduction to the Logic Programming Paradigms with an Eye
to Agents Design
Enrico Pontelli
(SHORT) Towards Exposing
Cyberstalkers in Online Social
Networks (p.161)
Jiamou Liu, Yingying Tao and
Quan Bai.
18:30 - 21:00
Banquet (Ballroom B) (18:30 - 21:00)
63
DAY 4
Timetable
THURSDAY, AUGUST 25
Ballroom B
Arcadia 1
08:30 - 09:00
Registration
09:00 - 10:00
From AdaBoost to Optimal Margin Distribution Machines
Zhi-Hua Zhou
(Keynote Speech)
(Ballroom B)
BREAK
10:00 - 10:30
10:30 - 12:00
(Session 4)
12:00 - 13:30
Reinforcement Learning
(Session Chair:Yang Yu)
Image Processing
(Session Chair: Andy Song)
A novel multi stage cooperative
path replanning method for
multi UAV (p.114)
Xiao-hong Su, Ming Zhao, Lingling Zhao and Yan-hang Zhang.
Generalized Extreme Value Filter to
Remove Mixed Gaussian-Impulse
Noise (p.135)
Sakon Chankhachon and Sathit
Intajag.
Faster Convergence to Cooperative Policy by Autonomous
Detection of Interference States
in Multiagent Reinforcement
Learning (p.134)
Sachiyo Arai and Haichi Xu.
3-D Volume of Interest Based Image
Classification (p.108)
Akadej Udomchaiporn, Frans
Coenen, Marta Garcia-Finana and
Vanessa Sluming.
Exploring Multi-Action Relationship in Reinforcement
Learning (p.131)
Han Wang and Yang Yu.
Single Image Super-Resolution
Based on Nonlocal Sparse and LowRank Regularization (p.157)
Chunhong Liu, Faming Fang, Yingying Xu and Chaomin Shen.
(SHORT) Hybrid Temporal-Difference Algorithm using Sliding
Mode Control and Sigmoid
Function (p.136)
Ke Xu and Fengge Wu.
(SHORT) Learning with additional
distributions (p.143)
Sanparith Marukatat.
LUNCH & POSTER
64
DAY 4
Timetable
THURSDAY, AUGUST 25
Arcadia 2
Ballroom Foyer
08:30 - 09:00
Registration
09:00 - 10:00
(Ballroom B)
Zhi-Hua Zhou
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 4)
PRIMA Tutorial 2
Poster sessions
PM-T 2
PRIMA-Tutorial 2
Mini-School
Agent technology and business
process management: A new
synthesis
Aditya Ghose
POSTER SESSION
(At Ballroom Foyer,
coffee break area)
(Place poster at 09:00-10:00)
(Present 10:00-12:30)
(10:00 - 13:00)
12:00 - 13:30
LUNCH & POSTER
65
DAY 4
THURSDAY, AUGUST 25
Timetable
Ballroom B
Arcadia 1
13:30 - 15:00
Natural Language Processing
(Session Chair: Thepchai Supnithi)
Pattern Recognition
(Session Chair: Young-Bin Kwon)
(Session 5)
Selecting Training Data for Unsupervised Domain Adaptation
in Word Sense Disambiguation
(p.153)
Kanako Komiya, Minoru Sasaki,
Hiroyuki Shinnou, Yoshiyuki Kotani and Manabu Okumura.
An Orientation Histogram based
Approach for Fall Detection Using
Wearable Sensors (p.123)
Diep Nguyen Ngoc, Cuong Pham and
Phuong Tu.
Learning from Numerous Untailored Summaries (p.142)
Yuta Kikuchi, Akihiko Watanabe,
Ryohei Sasano, Hiroya Takamura
and Manabu Okumura.
Motion Primitive Forests for
Human Activity Recognition using
Wearable Sensors (p.147)
Diep Nguyen Ngoc, Cuong Pham and
Phuong Tu.
Incorporating an Implicit and
Explicit Similarity Network for
User-level Sentiment Classification on Microblogging (p.137)
Yongyos Kaewpitakkun and Kiyoaki Shirai.
Acquiring Activities of People Engaged in Certain Occupations
(p.117)
Miho Matsunagi, Ryohei Sasano,
Hiroya Takamura and Manabu Okumura.
(SHORT) Learning Sentimental
Weights of Mixedgram Terms for
Classification and Visualization
(p.143)
Tszhang Guo, Bowen Li, Zihao
Fu, Tao Wan and Zengchang Qin.
(SHORT) Thai Printed Character
Recognition Using Long ShortTerm Memory and Vertical Component Shifting (p.159)
Taweesak Emsawas and
Boonserm Kijsirikul.
15:00 - 15:30
BREAK
66
DAY 4
Timetable
13:30 - 15:00
(Session 5)
THURSDAY, AUGUST 25
Arcadia 2
Ballroom Foyer
PRIMA Tutorial 3 (1)
-
PM-T 3/1
PRIMA-Tutorial 3/1
Mini-School
Market Design: Designing Social
System by Game Theory
Makoto Yokoo
15:00 - 15:30
BREAK
67
-
DAY 4
Timetable
15:30 - 17:00
(Session 6)
THURSDAY, AUGUST 25
Ballroom B
Arcadia 1
Textual Analysis
Computer Vision
(Session Chair: Manabu Okumura) (Session Chair: Sanparith Marukatat)
An FAQ Search Method Using
a Document Classifier Trained
with Automatically Generated
Training Data (p.121)
Takuya Makino, Tomoya Noro and
Tomoya Iwakura.
Set to Set Visual Tracking (p.155)
Wencheng Zhu, Pengfei Zhu, Qinghua
Hu and Changqing Zhang.
Sentiment Analysis for Images
on Microblogging by Integrating
Textual Information with Multiple Kernel Learning (p.154)
Junxin Tan, Mengting Xu, Lin
Shang and Xiuyi Jia.
Multi-Level Occupancy Grids for
Efficient Representation of 3D
Indoor Environments (p.148)
Yu Tian, Wanrong Huang, Yanzhen
Wang, Xiaodong Yi, Zhiyuan Wang
and Xuejun Yang.
Fast Training of a Graph
Boosting for Large-Scale Text
Classification (p.134)
Hiyori Yoshikawa and Tomoya
Iwakura.
Early Detection of Osteoarthritis
Using Local Binary Patterns: A
Study Directed at Human Joint
Imagery (p.130)
Kwankamon Dittakan and Frans
Coenen.
(SHORT) Grouped Text Clustering Using Non-Parametric
Gaussian Mixture Experts
(p.136)
Yong Tian, Yu Rong, Yuan Yao,
Weidong Liu and Jiaxing Song.
68
DAY 4
Timetable
15:30 - 17:00
(Session 6)
THURSDAY, AUGUST 25
Arcadia 2
Ballroom Foyer
PRIMA Tutorial 3 (2)
-
PM-T 3/2
PRIMA-Tutorial 3/2
Mini-School
Market Design: Designing Social
System by Game Theory
Makoto Yokoo
69
-
DAY 5
Timetable
Andaman
FRIDAY, AUGUST 26
Similan
08:30 - 09:00
Registration
09:00 - 10:00
(Arcadia Hall)
Argumentation for Practical Reasoning
Phan Minh Dung
Asian Institute of Technology, Thailand
[email protected]
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 7)
Search
(Session Chair: Katsutoshi Hirayama)
Data Mining
(Session Chair: Xin Geng)
Local Search with Noisy Strategy
for Minimum Vertex Cover in Massive Graphs (p.144)
Zongjie Ma, Yi Fan, Kaile Su,
Chengqian Li and Abdul Sattar.
Detecting Critical Links in
Complex Network to Maintain
Information Flow/Reachability
(p.128)
Kazumi Saito, Masahiro Kimura, Kouzou Ohara and Hiroshi
Motoda.
Combining Swarm with Gradient
Search for Maximum Margin Matrix Factorization (p.127)
Salman K H, Arun K Pujari, Vikas
Kumar and Sowmini Devi Veeramachaneni.
A Multi-Memory Multi-Population Memetic Algorithm for
Dynamic Shortest Path Routing
in Mobile Ad-hoc Networks
(p.112)
Nasser R. Sabar, Ayad Turky and
Andy Song.
(SHORT) Generating Covering
Arrays with pseudo-Boolean
Constraint Solving and Balancing
Heuristic (p.135)
Hai Liu, Feifei Ma and Jian Zhang.
SWARM: An Approach for Mining Semantic Association Rules
from Semantic Web Data
(p.158)
Molood Barati, Quan Bai and
Qing Liu.
(SHORT) An empirical local search
for the stable marriage problem
(p.120)
Hoang Huu Viet, Le Hong
Trang, Seunggwan Lee and Taechoong Chung.
(SHORT) An Investigation
of Objective Interestingness
Measures for Association Rule
Mining (p.122)
Rachasak Somyanonthanakul
and Thanaruk Theeramunkong.
12:00 - 13:30
LUNCH
13:30 - 17:30
Excursion (Need to register at the registration desk
within August 24, 2016)
70
DAY 5
Timetable
Arcadia 1
FRIDAY, AUGUST 26
Arcadia 2
08:30 - 09:00
Registration
09:00 - 10:00
(Arcadia Hall)
Phan Minh Dung
[email protected]
10:00 - 10:30
BREAK
(Keynote Speech)
10:30 - 12:00
(Session 7)
Biometric Recognition
(Session Chair: Pengfei Zhu)
PRIMA Tutorial 4
Facial Age Estimation by Total
Ordering Preserving Projection
(p.133)
Xiao-Dong Wang and Zhi-Hua
Zhou.
Face Verification Algorithm with
Exploiting Feature Distribution
(p.132)
Zhi Qu, Xuan Li and Yong Dou.
Large Margin Coupled Mapping
For Low Resolution Face Recognition (p.141)
Jiaqi Zhang, Zhenhua Guo, Xiu Li
and Youbin Chen.
PM-T4
PRIMA-Tutorial 4
Mini-School
Sequential Decision Making for
Improving Efficiency in Urban
Environments
Pradeep Varakantham
(10:00-13:00)
12:00 - 13:30
LUNCH
13:30 - 17:30
Excursion (Need to register at the registration desk
within August 24, 2016)
71
Conference Venue
72
Floor Plan (Andaman meeting room)
73
Floor Plan (Arcadia Hall)
74
Floor Plan (Grand Ballroom)
75
Floor Plan (Similan room)
76
Keynote/Invited Speakers
Two PRIMA Keynote/Invited Speakers
Toru Ishida
In the beginning of
the new millennium,
we proposed the concept of intercultural collaboration where participants with
different cultures and languages work
together towards shared goals. Because
intercultural collaboration is a new area
with scarce data, it was necessary to
execute parallel experiments in both in
real fields as well as in research laboratories. In 2002, we conducted a oneyear experiment with Japanese, Chinese, Korean and Malaysian colleagues
and students to develop open-source
software using machine translation.
From this experiment, we understood
the necessity of language infrastructure
on the Internet to create customized
multilingual environments for various situations. In 2006, we launched
the Language Grid project to realize a
federated operation of servers for language services. So far, four servers have
been set up in Asia, and more than 200
language services have been registered
from 22 countries. Using the Language
Grid, we have been working with an
international NGO for four years to
support communications between rice
harvesting experts in Japan and farmers
and their children in Vietnam. During
these experiences, we gradually understood the nature of intercultural collaboration and we faced different types
of difficulties. Problems are “wicked”
and not easily defined because of their
nested and open networked structure.
For example, technologies supporting
collaboration are often successful when
we weaken cultural differences; cognitions of cultural differences are different in different cultures, etc. In this
talk, we view intercultural collaboration
as a research in human-aware multiagent systems with the goal to encourage researchers in multiagent systems
to stand by intercultural collaboration.
We believe a human-aware multiagent research can contribute solutions to real-world complex problems.
77
Biography
Toru Ishida has been a professor of Kyoto University since 1993. He has
been a fellow of IEEE, a vice president of IEICE, and a member of the Science Council of Japan. He is a co-founder of the Department of Social Informatics, Kyoto University and the Kyoto University Design School. His
research interest lies with Autonomous Agents and Multi-Agent Systems
and modeling collaboration within human societies. He contributed to create PRIMA/ICMAS/AAMAS conferences: he was a chair of the first PRIMA, a program co-chair of the second ICMAS, and the general co-chair of
the first AAMAS. His projects include Community Computing, Digital City
Kyoto, Intercultural Collaboration Experiments, and the Language Grid.
Keynote: Monday, August 22 (09:00 - 10:00) Ballroom B
78
Two PRIMA Keynote/Invited Speakers
Agent-based modelling and simulation for co-operative traffic
and transport
Jörg P. Müller
Recent
developments in Car-to-X
communication networks, advanced
assistance functions, and autonous vehicles create new prospects for Intelligent Traffic and Transport Systems
(ITS): Traffic Information Systems
may benefit from real-time information provided via cooperative sensing;
autonomous vehicles can interact and
perform cooperative driving maneuvers, aiming at improving traffic safety, traffic efficiency, and user comfort;
traffic management may provide individual route guidance and electronic
road pricing mechanisms to influence
the behaviour of (automated and human) traffic participants towards societal goals. However, the autonomy
of traffic participants makes analysing, predicting, and managing future
ITS a very difficult problem, creating
challenges for next-generation traffic
management systems, requiring new
cooperative approaches, enabling us
to reconcile the concepts of user optimum and system optimum, respec-
tively. Starting from the notion of
co-operative, (de-)centralized traffic
management, I advocate the multiagent paradigm for modeling and simulation of future traffic systems. I report on experiences with a multi-agent
based approach for cooperative traffic
management and autonomous vehicle scenarios. I sketch the underlying
conceptual paradigm and architecture,
exemplify the use of game-theoretic
and computational social choice methods in the traffic application domain,
and report on ongoing work geared
towards platforms supporting scalable multi-agent based simulations.
79
Biography
Prof. Dr. Jörg P. Müller is a Full Professor of Computer Science at Clausthal
University of Technology; since 2008, he has been Head of the Department of
Informatics at TU Clausthal. Previously, Joerg was a senior researcher at Siemens AG Corporate Technology, where he led the agents and peer-to-peer computing research group; earlier employments include John Wiley & Sons, Zuno
Ltd., Mitsubishi Electric, and the German Artificial Intelligence Research Center
(DFKI). Joerg holds a Ph.D. from Saarbrucken University in 1996 and an M.Sc.
in Computer Science from Kaiserslautern University in 1991. Within the last 25
years, he has published over 200 papers on intelligent agents and multi-agent
systems, business information systems and distributed computing. His current
research interests include agent-based models, methods, technologies and applications for decentralized sociotechnical systems, which are developed and
validated in traffic/transport, automation, and product engineering applications.
Keynote: Tuesday, August 23 (09:00 - 10:00) Ballroom B
80
Two PRICAI Keynote/Invited Speakers
Sheng-Chuan Wu
Franz Inc., Silicon Valley, USA.
[email protected]
The explosion of data
may have made AI
trendy again. Google has just made
their AI chief the head of Google
Search, which holds the world’s biggest repository of data. One of the key
AI endeavors is knowledge acquisition and discovery. Typically, we turn
the data we collect into information by
applying its context. We then interpret
the information to derive knowledge
from it. Knowledge is what provides
value to our endeavors, as we believe. Is this paradigm still true with
the explosive growth in Big Data?
One of the most consequential examples is medical science. Since the
sequencing of the human genome in
2003, we have dreamed about treating patients more effectively based on
their genomic profiles. Such a dream
remains elusive. The fundamental
difficulty lies in the complexity of
biological systems that have evolved
through billions of years. On the other
hand, major progress can be and has
been made in “personalized medicine” by applying classic AI machine
learning on the massive patient medical data accumulated. In essence, we
can uncover new knowledge from the
data to help patients without knowing
the why a priori. Lack of direct value
brought forth the last AI winter. Perhaps Big Data will foreshadow the
coming spring of AI. Exploiting Big
Data brings another set of management problems, namely the heterogeneous nature of data sources and
taxonomies, the enormous size of
data volume, and huge data analytic
processing requirements. We will discuss all these issues and show some
examples in healthcare at this talk.
81
Biography
Dr. Sheng-Chuan Wu received his Ph.D. in Scientific Computing and Computer
Graphics from Cornell University in the US. He has, since graduation, involved
in several software companies, including the founding of the first integrated
CAD/CAM/CAE company. In the last 20 years, he worked as a senior corporate
executive at the leading Artificial Intelligence and Semantic Technology company, Franz Inc in Silicon Valley, with responsibility in application development,
marketing, consulting and new business development. Dr. Wu has also in many
occasions collaborated with Bioinformatics experts from Harvard Medical
School, Stanford University and Astra Zeneca, working with massive biological
data. Dr. Wu has been focusing on Semantic Technology over the last 8 years.
He routinely lectured on AI and Semantic Technology at conferences. He was
a keynote speaker at PRICAI 2004 in Auckland NZ. Most recently, he gave a
keynote at KSEM 2015 in China and will deliver another keynote at KMO 2016
in Germany. He has, since 2007, conducted more than 20 week-long workshops
on Semantic Technology and Artificial Intelligence in Malaysia, China, Singapore, India and other Asian countries. Dr. Wu has also consulted on several
Big Data and Semantic Technology projects in the US and Asia. Some of the
projects include: Biodiversity Repository, Precision Agriculture for Citrus Plantation, Telecom Customer Relation Management, Malaysia R&D Knowledgebase, Intelligence analytics, Meta Data Management and E-Learning System.
Keynote: Wednesday, August 24 (09:00 - 10:00) Ballroom B
82
Zhi-Hua Zhou
AdaBoost is a famous
mainstream ensemble
learning approach that has greatly influenced machine learning and related
areas. A well-known mystery of Adaboost lies in the phenomenon that it
seems resistant to overfitting, which
has inspired a lot of theoretical investigations. In this talk, we will briefly
introduce the margin theory that has a
long history of debating but recently
defensed. We will show how the theoretical findings provide inspiration for
Optimal margin Distribution Machines
(ODM), a promising direction of designing powerful learning algorithms
Biography
Zhi-Hua Zhou is a Professor and Founding Director of the LAMDA Group at
Nanjing University. He authored the book “Ensemble Methods: Foundations
and Algorithms”, and published more than 100 papers in top-tier journals and
conference proceedings. His work have received more than 20,000 citations,
with a h-index of 71. He also holds 14 patents and has good experiences in
industrial applications. He has received various awards, including the National Natural Science Award of China, the IEEE CIS Outstanding Early Career
Award, the Microsoft Professorship Award, 12 international journal/conference paper/presentation/competition awards, etc. He serves as the Executive
Editor-in-Chief of Frontiers of Computer Science, Associate Editor-in-Chief
of Science China, and Associate Editor of ACM TIST, IEEE TNNLS, etc.
He founded ACML (Asian Conference on Machine Learning) and served as
General Chair and Program Chairs for various conferences including PAKDD’07, PRICAI’08, SDM’13, ICDM’15, IJCAI’15 Machine Learning track,
etc. He also serves as Advisory Committee member for IJCAI 2015-2016,
and Steering Committee Member of PAKDD and PRICAI. He is a Fellow of
the AAAI, IEEE, IAPR, IET/IEE, CCF, and an ACM Distinguished Scientist
Keynote: Thursday, August 25 (09:00 - 10:00) Ballroom B.
83
One Co-PRICAI/PRIMA Keynote/Invited Speaker
Phan Minh Dung
Argumentation is a
key component in
human intellectual activities. We engage in argumentation in almost all
of our daily conversations. Much of
our knowledge and intellectual skills
is learned through argumentation. Can
we build humanoid robots that could
argue with us and help us improving
our own arguing skills? Argumentation is an active field in AI and knowledge representation. A well-known
model of an argumentation systems
(also known as argumentation frameworks) is a pair (AR,attack) where
AR is a set of arguments and att is a
binary relation over AR representing an attack relation between arguments. Though very simple, it turns
out that this model is general enough
to represent many forms of practical
reasoning like legal reasoning, common-sense or default reasoning, experimental reasoning (like in medicine). In this talk we will present an
axiomatic approach to the theory of
argumentation. We begin with a rather light-hearted illustration of the role
of argumentation in our daily lives
and proceed to present an overview
of the principles and properties underlining the theory of argumentation.
Biography
Currently I am a professor in computer science at AIT. Before joining AIT,
I have been working at the National Institute for Informatics in Hanoi Vietnam. I got my Master and PhD degrees at the Dresden University in Germany. The title of my PhD thesis is: Structure and Axioms of Nondeterministic Computation. I am an associate editor of the Journal of Artificial
Intelligence and an area editor of the Journal of Logic and Computation.
I am a member of the editorial board of the journals of Theory and Practice of Logic Programming and Journal of Argument and Computation.
Keynote: Friday, August 26 (09:00 - 10:00) Arcadia Hall
84
IWEC Invited Speaker
Recent advances in emotional and motivational circuits
Kenji TANAKA
Department of Neuropsychiatry,
Keio University School of Medicine.
psychological status in depth in humans. Last I will introduce the recent
technology, called optogenetics, and
will talk optogenetics-mediated approach toward deciphering emotional
and motivational circuits in animals.
What is emotion?
How can we realize
our emotional state? Why does emotional response coincide with physical
response? Why do depressive patients
suffer from decreased motivation?
It is difficult to clearly answer them,
however, there are several theories to
understand them based on the clinical/psychological observations. First
I will overview theories regarding
emotion. Second I will point out the
technical limitation to understand
Biography
Dr. Kenji Tanaka received his M.D. from Keio University School of Medicine,
Japan in 1997 and completed the residency training in the psychiatry department
of Keio University Hospital. He received his Ph.D. in Neuropsychiatry from Keio
University Graduate School of Medicine, Japan in 2003. After postdoctoral training at National Institute for Physiological Sciences, Japan (2003-2006, mentor:
Dr. Kazuhiro Ikenaka), Columbia University, USA (2006-2008, mentor: Dr. René
Hen), he appointed the assistant professor at National Institute for Physiological
Sciences in 2008. He is now the associated professor at Keio University School of
Medicine, Department of Neuropsychiatry. His research is aimed at understanding the mechanisms of recovery from and resilience to mood/anxiety disorders.
Keynote: Tuesday, August 23 (13:45 - 14:30) Arcadia 2
85
PRICAI Tutorials
Deep learning for natural language processing (T-1)
Assoc. Prof. Dr. Hidekazu Yanagimoto
College of Sustainable System Science, Osaka Prefecture University,
1-1, Gakuen-cho, Naka-ku, Sakai, Osaka, Japan
URL: http://www.sig.cs.osakafu-u.ac.jp/~hidekazu/
Instructor biographical data
Dr. Hidekazu Yanagimoto received B. Eng. degree, M. Eng. degree, and
Dr. Eng. degree from Osaka Prefecture University, Japan in 1994, 1996,
2006, respectively. He worked in NEC Laboratory from April, 1996 to
March, 2000. In April, 2000 he joined Osaka Prefecture University as research associate and currently as associate professor. He stayed Helsinki University of Technology as a visiting researcher from 2008 to 2009. His
research interests are Natural Language Processing and Machine Learning.
Tutorial Length
A six-hour tutorial
86
Objectives and motivation of the tutorial
Deep learning is one of the hottest research topics in artificial intelligence and
achieves goodresults in computer vision, speech recognition, and natural language processing. However, DeepLearning progresses rapidly and it is difficult
to understand overviews of deep learning. Hence,in this tutorial I will present a
brief introduction to deep learning from basic concepts, multilayered neural network in natural language processing. After the tutorial, audience can use deep
learning in their research areas.
Intended audience
1. Students/researchers who are interested in applying deep learning to
natural language processing
2. Students/researchers who are interested in deep learning
3. Students/researchers who are interested in machine learning.
Tutorial outline
A tutorial of deep learning for natural language processing from basic neural
network theory to
recent neural network systems with deep learning. The tutorial includes the
following topics
1. A basic multi-layer neural network
2. Back propagation as a Learning algorithm of the neural network
3. Gradient vanishing and explosion in a deep architecture neural network
4. Convolutional neural network
5. Recurrent neural network
6. Long short term memory
7. Encoder-Decoder model for machine translation
8. Attentional network for machine translation
9. Other neural networks (adversarial network, residual network, etc.)
Tutorial: Monday, August 22 (10:00 - 17:00) Similan
87
Introduction to Coalition Formation and its Application (T-2)
URL: https://khamreang.msu.ac.th/~chatra/
Dr. Sombattheera received his PhD in 2010 from University of Wollongong,
Australia. He received his Master ofInformation Technology in 1999 from
University of Sydney, and Graduate Diploma (Computer Science) in 1996
from University of Western Australia. He received his Bachelor of Science
(Computer Science) in 1992 from Ramkhamhaeng University, Thailand.
At present, Dr. Sombattheera teaches at Faculty of Informatics, Mahasarakham University, Thailand and leads the Multiagent, Intelligence and Simulation Laborlatory. In the last five years, he has been working on challenging projects with Defense Institute of Thailand and AVIA Sat Com
Ltd in attempts to bridge the gap between theoretical research and real
world problems. In addition, he also undertake non-military projects in
different areas, including composite services, logistics, intelligent web,
etc., which apply coalition formation concept to tackle the problems.
Tutorial Length
A four-hour tutorial
88
Coalition formation is an interesting area of research in multiagent systems,
an important area in artificialintelligence. Since 1994, when the first game
theorists, namely John Nash, Reinhard Selten and John Harsanyi, were rewarded the so called Nobel Prize in Economics (The Sveriges Riksbank Prize
in Economic Sciences in Memory of Alfred Nobel), AI communities have
been excited about challenges in solving hard problems in Economics, both
in terms of complexity and outcomes of non-cooperative games (defined by
Neumann and Morgenstern). However, the other types of games, cooperative
game or coalition formation is also interesting – there are cooperative game
theorists, namely Robert Aumann and Lloyd Shapley, awarded the same
prizes in 2005 and 2012 respectively. Coalition formation offers several
solution concepts, including Shapley value, Kernel, Core, which can be applied to solve computationally hard problems in real world, including optimization, logistics, combinatorial auction, mechanism design, etc. Although coalition formation is well-known internationally, it is hardly known
to Thai AI community and many young researchers. It is a good opportunity for them to disclose this exciting research area. The objective of this tutorial is to give an introductory lecture to coalition formation in AI research.
Intended audience
Master and PhD students in Computer Science (and interested people).
Tutorial outline
1. Introduction to game theory
2. Non-cooperative game
3. Coalition formation
4. Shapley value
5. Kernel
6. Core
7. Optimal Coalition Structure
8. Application of coalition formation Combinatorial auction, Logistics,
Mechanism design
Tutorial: Tuesday, August 23 (10:30 - 12:00) Arcadia 2
89
Collective Intelligence: Using Systems/Design Thinking Methods to Improve Group Intelligence (T-3)
Jarun Ngamvirojcharoen
Komes Chandavimol
Data Science Thailand and Sertis Corporation
Email: [email protected] and [email protected]
URL: http://www.linkedin.com/in/jarun
URL: https://www.linkedin.com/in/komescha
1. Jarun Ngamvirojcharoen
Jarun recently joined Sertis Corporation as a Chief Data Scientist/Vice President of DataInnovation Lab. In his role at Sertis, he is building a data science
team and is looking for Artificial Intelligence & Machine Learning research collaboration opportunities with universities in the Asian region. Prior to Sertis, he
was a lead data scientist of Booz Allen Hamilton working in the area of fraud &
ID Theft detection and predictive modeling. He has been developing a holistic
framework incorporating data science, systems thinking, and design thinking for
creative problem solving. He has been a guest speaker to some leading institutes
in Thailand (e.g. NECTEC, Chulalongkorn University). Prior to Booz Allen, he
was a Founder/CEO of a startup Crono building an Artificial Intelligence (AI)
personal assistance for mobile services funded by an angel investor & National
Science Foundation (NSF). He spent over 15 years in software, telecommunications, and financial services industries playing a number of different roles in
management, system design, analytics, software development, and technology
startup consulting. Educational highlights include MBA from Carnegie Mellon University, a Master of Science in Telecommunications from University of
Pittsburgh, and Bachelor of Engineering in Electrical engineering from Chulalongkorn University. He is a Certified Professional InnovatorPractitioner-Design Thinking and a Certified Systemic Design Thinking and Facilitation.
90
2. Komes Chandavimo
Komes is a Lead Data Scientist at the Data Science Lab, Thailand. This lab
is part of the Data Science Community, where data professionals learn and
share Data Science experience for the benefit of Data Science in Thailand. His
lab is a place that academics meet practitioner and develop the practical data
science products. Komes is currently lead concurrent data science projects
(Proof-of-Concept) includes Market Basket Analysis, Propensity Model, Customer Segmentation, Sentimental Analysis, Real-time Steaming Analysis etc.
In the US, Komes has worked with several healthcare clients to implement BI/
Reporting systems. His clients include Kaiser Permanente, Cedar Sinai, University of Chicago Medical, Seattle Children Hospital, Yale health and Providence Health etc. His responsibilities not only focus on the system implementation but also the resource planning and team building. Lastly, Komes is
also working on his Doctoral of Business Administration at University of Liverpool. His thesis is to improve virtual team performance by self-leadership
and shared leadership. His thesis focus on action learning, which encourage
team members to reflect on their thinking process as well as evaluate their actions. This action learning process is not only improve the team performance
but also enhance individual productivity and self-learning as well.
Tutorial Length
A three-hour tutorial
91
In general problem solving, we are typically dealing with the incomplete information of a problem (wicked problem) and dispersed knowledge in our
working community. How do we collectively harness human intelligence of
a team in order to gain insight and knowledge of this complex problem? “Everyone designs who devises courses of action aimed at changing existing situations into preferred ones” In product and service design, methods in systems
thinking and design thinking have proven to be effective processes for team
collaboration to synthesize new knowledge. By utilizing different perspectives
and thinking processes, the team could act as a learning system of multiple
intelligent agents comprising of facilitator & governing mechanism working
together to solve problems. In this workshop, we will introduce techniques for
problem framing which is the most essential part of a problem solving process.
Intended audience
People who are interested in applying systems thinking and design thinking
techniques in their collaboration for problem framing and improving shared
understanding & team intelligence.
Tutorial outline
1. Introduction to Systemic Design Thinking
2. Frame-storming process
3. 5 Whys
4. Causal Analysis
5. Metamap (if time permitted)
Tutorial: Tuesday, August 23 (10:30 - 12:00) Similan
92
Visually See Text Mining Math Processes on LSA, SVD, and
Gibbs Sampling (T-4)
Prof. Yukari Shirota
Basabi Chakraborty, Prof. and Ph. D
Basabi Chakraborty, Prof. and Ph. D
Gakushuin University, Japan
1. Yukari Shirota
DSc. Prof. of Faculty of Economics, Gakushuin University. Research fields are
visualization of data on the web, data visualization, social media analysis, and
visual education methods for business mathematics. For over 17 years, she has
developed visual teaching materials for business mathematics and statistics.
93
2. Basabi Chakraborty, Prof. and Ph. D
She received B.Tech, M.Tech and Ph. D degrees in Radio Physics and Electronics from Calcutta University, India and worked in Indian Statistical Institute, Calcutta, India until 1990. She received another Ph. D in Information Science from Tohoku University, Japan in 1996. Currently she is a full
professor in Software and Information Science department of Iwate Prefectural University, Japan. Her main research interests are in the area of Pattern Recognition, Machine Learning, Soft Computing Techniques, Data
mining and Online Social media mining. She is a senior member of IEEE,
member of ACM, INNS and Japanese Society of Artificial Intelligence.
3. Takako Hashimoto, Prof. and Ph. D.
She received a Ph.D. in computer science, specialization in multimedia information processing, from the Graduate School of Systems and Information
Engineering of University of Tsukuba in 2005. From April of 2009, she was
involved in Chiba University of Commerce as Associate Professor. In 2015,
she has become Professor of Chiba University of Commerce and stayed at
University of California, Los Angeles as a visiting researcher. From 2016,
she is Director of Economic Research at Chiba University of Commerce.
She has focused on the data mining research and the social media analysis,
especially topic extraction from millions of tweets related to the East Japan
Great Earthquake. She is developing the high performance feature selection
technique for big data. She’s also conducting global researches for developing the social media analysis platform in multi-language/cultural environment
Tutorial Length
A three-hour tutorial
94
Visualization has the power to make people understand the content meaning at a
glance and the visualization power is required also in mathematics education. We
have developed and used visual teaching materials in our university math lectures. We would like to show our teaching materials and explain the mathematics
visually and interactively. The materials have been already published on web.
Therefore the audience can in advance access and operate the visual materials
with the free software “Wolfram CDF player.” We hope that more people who
need and studymathematics easily understand and enjoy mathematics visually.
Intended audience
1. People who use text mining programmers for their works but feel difficul-
ties in studying the mathematics themselves.
2. Text mining lecturers who would like to see the visual approach for teaching the mathematics/statistics
Tutorial outline
1. Singular Value Decomposition (SVD) in Latent Semantic Analysis (LSA)
2. Singular Value Decomposition in Time Series Data Analysis
3. Primary Component Analysis (PCA)
4. Gibbs Sampling
Tutorial: Tuesday, August 23 (13:30 - 17:00) Similan
95
PRIMA Tutorials (Mini-School)
An Introduction to the Logic Programming Paradigms with
an Eye to Agents Design (PM-T 1)
Prof. Enrico Pontelli
Computer Science Department
New Mexico State University, United States
URL: http://www.cs.nmsu.edu/wp13/enrico-pontelli/
Abstract
The tutorial will provide a brief tour-de-force through the three main “flavors” of logic programming paradigms - traditional Prolog, constraint logic
programming, and answer set programming. We will review the foundations
of the three paradigms, the typical execution models, and implementations.
We will draw illustrative examples drawn from the use of these paradigms in
implementing reasoning components of intelligent knowledge-based agents.
96
Biography
Enrico Pontelli received a Laurea in Computer Science from the University of
Udine (1991), a Master’s from the University of Houston (1992) and a Ph.D.
from New Mexico State University (1997). He joined the faculty at NMSU in
1997, achieving the rank of Regents Professor. He is currently serving as Interim
Dean of the College of Arts and Sciences. His research interests are in the areas
of logic programming, constraint programming, knowledge representation and
reasoning, high performance computing, bioinformatics and assistive technologies. He has served as lead investigator for funded research projects for over
$15M, including a NSF Career award. He was recognized as a Distinguished
Achievement faculty by NMSU in 2012. He served in many national and international organization, including the Association of Logic Programming, the ACM
Special Interest Group on Computers and Accessibility, and the Computing
Research Association. He has published over 250 peer-reviewed publications.
Tutorial: Wednesday, August 24 (13:30 - 17:30) Room Arcadia 2
97
Agent technology and business process management: A new
synthesis (PM-T 2)
Prof. Aditya Ghose
School of Computing and Information Technology
URL: http://www.uow.edu.au/~aditya/
Abstract
The field of business process management is generally viewed as being distinct
to the field of agent technology. However, a closer analysis reveals considerable
similarity in their objectives and in the key units of analysis and representational
constructs. Developments in both fields are also complementary in many ways,
and each can leverage results from the other. This tutorial will lay out a new
synthesis of both approaches. Attendees will take away an appreciation of how
agent systems insights can add value in the (already substantial) industrial applications of business process management and how business process insights
might make agent systems more amenable for deployment in business/industry
settings. The tutorial will also highlight a number of research challenges where
the cross-fertilization of ideas from both fields can lead to interesting results.
98
Biography
Aditya Ghose is a Professor of Computer Science at the University of Wollongong, where he heads the Decision Systems Lab and co-directs the Centre for Oncology Informatics. He holds a PhD and MSc in Computing Science from the University of Alberta (he spent parts of his candidature at the
University of Illinois at Urbana-Champaign and the University of Tokyo)
and a Bachelor of Computer Science and Engineering from Jadavpur University. Prof. Ghose is President of the Service Science Society of Australia
and a Fellow of the Institution of Engineers (Australia). Prof. Ghose works
in business process management, service science, agent systems and formal
knowledge representation and the application of these in clinical informatics.
He works closely with IBM Research, Xerox Research, Infosys Labs, a number of radiation oncology centres and medical research institutes as well as a
variety of Australian companies (including startups from his research group).
Tutorial: Thursday, August 25 (10:00 - 13:00) Room Arcadia 2
99
Market Design: Designing Social System by Game Theory
(PM-T 3)
Prof. Makoto Yokoo
Department of Informatics
Graduate School of ISEE, Kyushu University, Japan
URL: http://agent.inf.kyushu-u.ac.jp/~yokoo/
Abstract
“Market Design” is a research field that examines how to design a new market or improve an existing market such that a certain design goal is satisfied.
This research field is influenced by micro economics, in particular, game theory. Here, the meaning of a “market” is very broad; it includes a spectrum
auction, in which a government allocates the licenses to use specific spectrum bandwidth to companies, or a market without monetary transfer, such as
a kidney exchange program, or a school choice program, in which children/
parents can choose public schools they want to attend. Now, market design
has become a interdisciplinary research topic that is relevant to computer science and information systems. In this lecture, I will describe two representative application domains of market design: auction mechanisms
(e.g., spectrum auctions, sponsored search) and two-side matching mechanisms (e.g., medical resident matching programs, school choice programs).
100
Biography
Makoto Yokoo received the B.E., M.E., and Ph.D. degrees in 1984, 1986, and
1995, respectively, form the University of Tokyo, Japan. He is currently a Distinguished Professor of Information Science and Electrical Engineering, Kyushu
University, Japan. He served as a general co-chair of International Conference
on Autonomous Agents and Multi-Agent Systems in 2007 (AAMAS-2007), and
as a program co-chair of AAMAS-2003. He served as the president of International Foundation for Autonomous Agent and Multiagent Systems (IFAAMAS)
from 2011 to 2013. He is a fellow of the Association for Advancement of Artificial Intelligence (AAAI). He received the ACM SIGART Autonomous Agents
Research Award in 2004, and the IFAAMAS influential paper award in 2010.
Tutorial: Thursday, August 25 (13:30 - 17:00) Room Arcadia 2
101
Sequential Decision Making for Improving Efficiency in
Urban Environments (PM-T 4)
Prof. Pradeep Varakantham
School of Information Systems
Singapore Management University, Singapore
URL: http://www.mysmu.edu/faculty/pradeepv/
Abstract
Rapid “urbanization” (more than 50% of worlds’ population now resides in cities)
coupled with the natural lack of coordination in usage of common resources (ex:
bikes, ambulances, taxis, traffic personnel, attractions) has a detrimental effect
on a wide variety of response (ex: waiting times, response time for emergency
needs) and coverage metrics (ex: predictability of traffic/security patrols) in cities of today. Motivated by the need to improve response and coverage metrics in
urban environments, my research group is focussed on building intelligent agent
systems that make sequential decisions to continuously match available supply
of resources to an uncertain demand for resources. Our broad approach to generating these sequential decision strategies is through a combination of data analytics (to obtain a model) and multi- stage optimization (planning/scheduling)
under uncertainty (to solve the model). While we perform data analytics, our
contributions are focussed on multi-stage optimization under uncertainty. We exploit key properties of urban environments, namely homogeneity and anonymity,
limited influence of individual entities, abstraction and near decomposability to
solve ”multi-stage optimization under uncertainty” effectively and efficiently.
102
Biography
Pradeep Varakantham is a Lee Kong Chian Fellow and an Assistant Professor in the School of Information Systems at Singapore Management University. Prior to his current position, Pradeep received his PhD from University
of Southern California and was a post-doctoral fellow for two years at Carnegie Mellon University. His research is at the intersection of Artificial Intelligence, Operations Research and Machine Learning with specific focus
on solving sequential matching problems in urban environments. Pradeep
has published over 60 research papers in top tier conferences (AAAI, IJCAI, AAMAS, ICAPS, UAI, NIPS) and journals (JAIR, JAAMAS) in Artificial Intelligence and Machine Learning. One of his papers was nominated
for best student paper award at AAMAS 2009 and he currently serves on the
board of directors for IFAAMAS (governing body of AAMAS). He was finalist for the best senior program committee member award at AAMAS 2013.
Tutorial: Friday, August 26 (10:00 - 13:00) Room Arcadia 2
103
Special Track (SMS)
Special Track on Smart Modelling and Simulation (SMS)
Track Chairs:
Dr. Quan Bai
Auckland University of Technology, New Zealand
Prof. Minjie Zhang
Prof. Takayuki Ito
Dr. Fenghui Ren
A/Prof. Katsuhide Fujita
Scope and Background
Computer based modelling and simulation has become useful tools to facilitate humans to understand systems in different domains, such as physics, astrophysics, chemistry, biology, economics, engineering and social science. A complex system is featured with a large number of interacting components (agents,
processes, etc.), whose aggregate activities are nonlinear and self-organized.
Complex systems are hard to be simulated or modelled by using traditional
computational approaches due to the complex relationships of components and
distributed features of resources, and dynamic work environments. Meanwhile,
smart systems such as multi-agent systems have demonstrated advantages and
great potentials in modelling and simulating complex systems. We are going
to organize a special session on Smart Modelling and Simulation (SMS) under
the 14th Pacific Rim International Conference on Artificial Intelligence (http://
saki.siit.tu.ac.th/pricai2016/). SMS aims to bring together researchers in both
artificial intelligence and system modelling/simulation to discuss research chal104
lenges and cutting edge techniques in smart simulation and modelling. Smart
simulation and modelling is an important field in AI. We believe that this special
session will attract attentions from not only AI researchers, but also domain experts who are interested in the applications of AI techniques in system modelling
and simulation. The topics of the special session include but are not limited to:
· Agent based simulation for complex systems
· Agent based modelling for complex systems
· Large-scale simulations
· Network simulation and modelling
· Environment and ecosystem simulation and modelling
· Smart Grid/Service simulation and modelling
· Simulation of social and economic organizations
· Simulation of social complexity
· Cooperation, coordination, negotiation and self-organisation in
complex systems
· Market-based model and simulation
· Transportation model and simulation
· Crowd model and simulation
· Evacuation model and simulation
· Simulation and modelling for disaster management
· Simulation and modelling techniques in Big Data
Special Session: Wednesday, August 24 (13:30 - 17:00) Room Arcadia 1
105
Abstracts
106
PRICAI
Abstracts
107
3-D Volume of Interest Based Image Classication
Akadej Udomchaiporn1, Frans Coenen1, Marta Garca-Fi~nana2,
and Vanessa Sluming3
1Department of Computer Science, University of Liverpool, Liverpool, UK
{akadej,coenen}@liv.ac.uk
2Department of Biostatistics, University of Liverpool, Liverpool, UK
[email protected]
3School of Health Science, University of Liverpool, Liverpool, UK
[email protected]
Abstract. This paper proposes a number of techniques for 3-D image classication
according to the nature of a particular Volume of Interest (VOI) that appears across
a given image set. Three VOI Based Image Classication (VOIBIC) approaches are
considered: (i) Statistical metric based, (ii) Point series based and (iii) Tree based.
For evaluation purpose, two 3-D MRI brain scan datasets, Epilepsy and Musicians,
were used; the aim being to distinguish between: (i) epilepsy patients versus healthy
people and (ii) musicians versus non-musicians. The paper also considers augmenting the VOI data with meta data. According to the reported experimental results
the Point series based approach, augmented with meta data, is the most effective.
Keywords: Image mining, Image classication, 3-D Magnetic Resonance Imaging (MRI)
108
A Concurrent Multiple Negotiation Mechanism under Consideration of A
Dynamic Negotiation Environment
Lei Niu, Fenghui Ren and Minjie Zhang
University of Wollongong, Wollongong, NSW, Australia
[email protected], {fren,minjie}@uow.edu.au
Abstract. Concurrent Multiple Negotiation (CMN) mechanism is necessary for agents
to achieve agreements in multiple negotiations, and it has become a very important
research topic in multi-agent systems in recent years. However, in the open and dynamic negotiation enviroment, negotiations may be dynamically and concurrently initialized or terminated during the process of other existing negotiations. Therefore, how
to process dynamic CMN becomes a serious challenge in agent negotiation research.
The motivation of this paper is to propose an adaptive mechanism for handling dynamic CMN by considering the possible changes of concurrent negotiations. First, a
formal mechanism for modeling and representing dynamic CMN is presented. Then,
a novel Colored Petri Net-based CMN protocol for processing CMN with unexpected
negotiation changes is presented. We also demonstrate the performance and procedure
of the proposed approach in handling the dynamism of negotiations in CMN, and the
experimental results show that the proposed approach can effectively handle unexpected changes in the CMN dynamically, and successfully lead the CMN to agreements.
109
A Differential Evolution Approach to Feature Selection
and Instance Selection
Jiaheng Wang, Bing Xue*, Xiaoying Gao, and Mengjie Zhang
School of Engineering and Computer Science
Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand
[email protected]
Abstract. More and more data is being collected due to constant improvements
in storage hardware and data collection techniques. The incoming flow of data is
so much that data mining techniques cannot keep up with. The data collected often
has redundant or irrelevant features/instances that limit classification performance.
Feature selection and instance selection are processes that help reduce this problem
by eliminating useless data. This paper develops a set of algorithms using Differential Evolution to achieve feature selection, instance selection, and combined feature
and instance selection. The reduction of the data, the classification accuracy and the
training time are compared with the original data and existing algorithms. Experiments on ten datasets of varying difficulty show that the newly developed algorithms
can successfully reduce the size of the data, and maintain or increase the classification
performance in most cases. In addition, the computational time is also substantially
reduced. This work is the first time for systematically investigating a series of algorithms on feature and/or instance selection in classification and the findings show that
instance selection is a much harder task to solve than feature selection, but with effective methods, it can significantly reduce the size of the data and provide great benefit.
Keywords: Differential Evolution; Feature Selection; Instance Selection; Classification
110
A Microblog Hot Topic Detection Algorithm based on
Discrete Particle Swarm Optimization
Huifang Ma, Yugang Ji, Xiaohong Li, Runan Zhou
College of Computer science and engineering, Northwest Normal University
[email protected]
Abstract. Traditional hot topic detection algorithms cannot show its optimal performance on microblogs for their inherent flaws in constructing short-text representation
model, implementing the core algorithm in large corpus with short time and evaluating the algorithms’ qualities during the process of detecting hot topics. In this paper, a novel method for detecting hot topics in microblogs is presented. This approach
takes advantage of a probabilistic correlation-based representation measure in order to ensure a dense and low-dimension microblog representation matrix. Besides,
we take the clustering as an optimization problem and introduce a discrete particle
swarm optimization (DPSO) to simplify the clustering process to detect topics. Furthermore, the clustering quality evaluation criteria is adopted as the optimization objective function for topic detection which can evaluate the algorithms’ qualities after
each iteration. Experimental results with corpora containing more than 148,000 twitters show that our algorithm is an effective hot topic detection method for microblog.
Keywords: Microblog; Hot Topic Detection; Probabilistic Correlation-based Representation; Short-Text Representation Model; Discrete Particle Swarm Optimization
111
A Multi-Memory Multi-Population Memetic Algorithm for Dynamic
Shortest Path Routing in Mobile Ad-hoc Networks
Nasser R. Sabar, Ayad Turky and Andy Song
School of Computer Science and I.T., RMIT University, Australia
{nasser.sabar, ayad.turky, andy.song}@rmit.edu.au
Abstract. This study investigates the dynamic shortest path routing (DSPR) problem
in mobile ad-hoc networks. The goal is to nd the shortest possible path that connects a
source node with the destination node while effectively handling dynamic changes occurring on the ad-hoc networks. The key challenge in DSPR is how to simultaneously
keep track changes and search for the global optima. A multi-memory based multi-population memetic algorithm is proposed for DSPR in this paper. The proposed algorithm
combines the strength of three dierent strategies, multi-memory and multi-population
and memetic algorithm, aiming to effectively explore and exploit the search space.
It divides the search space by multiple populations. The distribution of solutions in
each population is kept in the associated memory. The multi-memory multi-population approach is to capture dynamic changes and maintain search diversity. The memetic component, which is a hybrid Genetic Algorithm (GA) and local search, is to
nd high quality solutions. The performance of the proposed algorithm is evaluated on
benchmark DSPR instances under both cyclic and acyclic environments. Our method
obtained better results when compared with existing methods in the literatures, showing the effectiveness of the proposed algorithm in handling dynamic optimisation.
Keywords: dynamic shortest path routing; memetic algorithms; dynamic optimisatio;
evolutionary algorithm
112
A Novel Isolation-Based Outlier Detection Method
Yanhui Shen, Huawen Liu, Yanxia Wang*, Zhongyu Chen, Guanghua Sun
Department of Computer Science, Zhejiang Normal University, Jinhua, 321004, China
[email protected]
Abstract. Outlier detection is one of the most important tasks in data analysis. It
refers to the process of recognizing unusual characteristics which may provide useful insights in helping us to understand the behaviors of data. In the pa-per, an isolation-based outlier detection method, called Entropy-based Greedy Isolation Tree
(EGiTree), is proposed. Unlike other tree-like detection methods, our method exploits
a half-baked isolation tree, which is constructed via three entropy-based heuristics,
to identify outliers. Specifically, the heuristics are used to guide the selection process
of attribute and its split value when constructing the tree. Thus, the outlierness score
of each data point is estimated based on the total partition cost of the isolation node
in the tree, as well as the path length and complexity of partition. Experiment results
on public real-world datasets show that our approach outperforms distanced-based,
density-based, subspace-based as well as state of-the-art isolation-based approaches.
Keywords: Outlier detection, data mining, isolation, isolation tree, entropy
113
A Novel Multi Stage Cooperative Path
Re-planning Method for Multi UAV
Xiao-hong Su, Ming Zhao, Ling-ling Zhao, and Yan-hang Zhang
School of Computer Science and Technology, Harbin Institute of Technology, 150001
Harbin, China
Abstract. When the multi-UAVs cooperatively attack multi-tasks, the dynamic
changes of environments can lead to a failure of the tasks. So a novel path re-planning algorithm of multiple Q-learning based on cooperative fuzzy C means clustering is proposed. Our approach first reflects the dynamic changes of re-planning space
by updating the fuzzy cooperative matrix. Then, the key way-points on the current
global paths are used as the initial clustering centers for the cooperative fuzzy C
means clustering, which generates the classifications of space points for multitasks.
Furthermore, we use the classifications as the state space of each task and the fuzzy
cooperative matrix as the reward function of the Q-learning. So a multi Q-learning
algorithm is presented to synchronously re-plan the paths for multi-UAVs at every step. The simulation results show that the method subtracts the re-planning
space of the tasks and improves the search efficiency of the learning algorithm,
Keywords: Multi-UAVs; cooperative fuzzy C means clustering; multi Q-learning;
path re-planning.
114
A Relaxed K-SVD Algorithm for Spontaneous
Micro-Expression Recognition
Hao Zheng123 Xin Geng2;Zhongxue Yang1
1Key Laboratory of Trusted Cloud Computing and Big Data Analysis, School of
Information Engineering, Nanjing XiaoZhuang University, Nanjing, China
2MOE Key Laboratory of Computer Network and Information Integration, School of
Computer Science and Engineering, Southeast University, Nanjing, China
3State Key Laboratory for Novel Software Technology,
Nanjing University, Nanjing, China
Abstract. Micro-expression recognition has been a challenging problem in computer vision due to its subtlety, which are often hard to be concealed. In the paper,
a relaxed K-SVD algorithm (RK-SVD) to learn sparse dictionary for spontaneous
micro-expression recognition is proposed. In RK-SVD, the reconstruction error
and the classication error are considered, while the variance of sparse coefficients is
minimized to address the similarity of same classes and the distinctiveness of dierent classes. The optimization is implemented by the K-SVD algorithm and stochastic
gradient descent algorithm. Finally a single overcomplete dictionary and an optimal
linear classier are learned simultaneously. Experimental results on two spontaneous
micro-expression databases, namely CASME and CASME II, show that the performance of the new proposed algorithm is superior to other state-of-the-art algorithms.
Keywords: Related K-SVD, Dictionary learning, Micro-expression recognition
115
A Study of Players’ Experiences during Brain Games Play
Faizan Ahmad123, Yiqiang Chen1, Shuangquan Wang1, Zhenyu Chen12, Jianfei
Shen12, Lisha Hu12, Jindong Wang12
1Beijing Key Laboratory of Mobile Computing and Pervasive Device, Institute of
Computing Technology, Chinese Academy of Sciences
2University of Chinese Academy of Sciences, Beijing, China
3COMSATS Institute of Information Technology, Lahore, Pakistan
[email protected], {yqchen, wangshuangquan, chenzhenyu, shenjianfei,
hulisha, wangjindong}@ict.ac.cn
Abstract. Much of the experience of videogame players remains hidden. This paper presents an empirical study that assesses the experience of 50 participants (i.e.
25 children and 25 adults) during brain games play. Results from the empirical study
show a number of significant correlations among diverse kinds of players’ experiences (i.e. engagement, enjoyment, anxiety, usability, adaptability and noninvasiveness).
It is further identified by the study that the similarities and differences exist among
the experiences of children and adults. Consequently, the observations of presenting
study provide an insight against the experience of players during brain games play,
which was previously unknown. Besides, we exploit these insights to successfully narrow down the complexity of user feedback process for brain games playing activity.
Keywords: gamification, experiences, smart assessment, children, adults.
116
Acquiring Activities of People Engaged in Certain Occupations
Miho Matsunagi, Ryohei Sasano, Hiroya Takamura, and Manabu Okumura
[email protected]
{sasano,takamura,oku}@pi.titech.ac.jp
Abstract. We present a system to acquire knowledge on the activities of people engaged in certain occupations. While most of the previous studies acquire phrases related to the occupation, our system acquires pairs of a verb and one of its arguments, which we call activities. Our system acquires activities from sentences
written by people engaged in the target occupations as well as from sentences
whose subjects are the target occupations. Through experiments, we show that the
activities collected from each resource have dierent characteristics and the system based on the two resources would perform robustly for various occupations.
Keywords: activity acquisition, occupational knowledge, social media
Adaptive Learning for Efficient Emergence of Social Norms in Networked
Multiagent Systems
Chao Yu1, Hongtao Lv1, Sandip Sen2, Fenghui Ren3, and Guozhen Tan*1
1School of Computer Science and Technology
Dalian University of Technology, Dalian, 116024, China
2Department of Mathematical & Computer Sciences
University of Tulsa, Tulsa, Oklahoma 74104, United States
3School of Computer Science and Software Engineering
University of Wollongong, Wollongong, 2500, Australia
[email protected],[email protected],[email protected],
[email protected],[email protected]
Abstract. This paper investigates how norm emergence can be facilitated by agents’
adaptive learning behaviors in networked multiagent systems. A general learning framework is proposed, in which agents candynamically adapt their learning behaviors through
social learning of their individual learning experience. Extensive verication of the proposed framework is conducted in a variety of situations, using comprehensive evaluation
criteria of efficiency, effectiveness and efficacy. Experimental results show that the adaptive learning framework is robust and efficient for evolving stable norms among agents.
Keywords: Norm Emergence, Learning, Multiagent Systems
117
ALLIANCE-ROS: A Software Architecture on ROS for Fault-Tolerant
Cooperative Multi-Robot Systems
Minglong Li, Zhongxuan Cai, Xiaodong Yi, Zhiyuan Wang, Yanzhen
Wang,Yongjun Zhang, and Xuejun Yang
State Key Laboratory of High Performance Computing, College of Computer
National University of Defense Technology
Changsha, P. R. China, 410073
[email protected]
micros.trustie.net
Abstract. Programming multi-robot systems is a complicated and time-consuming work, due to two challenges, i.e., the distributed multi-robot cooperation and
the robot software reusability. ALLIANCE[1] is a fully distributed, fault-tolerant and behavior-based model. ROS (Robot Operating System) provides abundant
robot software modules. In this paper, by combining both, we propose a software architecture named ALLIANCE-ROS for developing fault-tolerant cooperativemulti-robot systems with a lot of software resources available.We encapsulate the ROS mechanisms and Python libraries to construct the basic function units of the ALLIANCE
model. One may inherit them to construct the ALLIANCE-modelapplication with
all ROS algorithms, modules and resources available. This work is demonstrated by
an experiment of multi-robot patrol in both the simulated and the real environments.
Keywords: multi-robot, fault-tolerant, cooperative, ALLIANCE-ROS
118
An Analysis of Influential Users for Predicting the Popularity of News
Tweets
Krissada Maleewong
School of Information Technology, Shinawatra University, Pathumthani, Thailand
[email protected]
Abstract. Twitter plays an important role in today social network. Its key mechanism is
retweet that disseminates information to broad audiences within a very short time and
help increases the popularity of the social content. Therefore, an effective model for
predicting the popularity of tweets is required in various domains such as news propagation, viral marketing, personalized message recommendation, and trend analysis.
Although many studies have been extensively researched on predicting the popularity of tweets, they mainly focus on the content-based and the author-based features,
while retweeter-based features are less concerned. This paper aims to study the impact of influential users who retweet tweets, also called retweeters, and presents simple yet effective measures for predicting the influence of retweeters on the popularity of online news tweets. By analyzing the popularity of news tweets and the impact
of the retweeters, a number of useful measures are defined to evaluate influence of
users in the retweeter network, and used to establish the prediction model. The experimental results show that the application of the retweeter-based features is highly
effective and enhances the performance of the prediction model with high accuracy.
Keywords: Twitter, Retweet, Influential User, Active User, Popular User, News
Tweet, Social Network
119
An Empirical Local Search for the Stable Marriage Problem
Hoang Huu Viet1, Le Hong Trang1, SeungGwan Lee2, and TaeChoong Chung3
1Department of Information Technology, Vinh University,
182-Le Duan, Vinh City, Nghe An, VietNam
([email protected](B), [email protected])
2Humanitas College, Kyung Hee University
3Department of Computer Engineering, Kyung Hee University,
1732, Deogyeong-daero, Giheung-gu, Yongin-si, Gyeonggi-do, 446-701, Korea
([email protected], [email protected])
Abstract. This paper proposes a local search algorithm to find the egalitarian and the
sex-equal stable matchings in the stable marriage problem. Based on the dominance
relation of stable matchings from the men’s point of view, our approach discovers the
egalitarian and the sex-equal stable matchings from the man-optimal stable matching. By employing a breakmarriage strategy to find stable neighbor matchings of the
current stable matching and moving to the best neighbor matching, our local search
nds the solutions while moving towards the woman-optimal stable matching. Simulations show that our proposed algorithm is efficient for the stable marriage problem.
120
An FAQ Search Method Using a Document Classifier Trained with Automatically Generated Training Data
Takuya Makino, Tomoya Noro, and Tomoya Iwakura
Fujitsu Laboratories Ltd.
{makino.takuya, t.noro iwakura.tomoya}@jp.fujitsu.com
Abstract. We propose an FAQ (Frequently Asked Question) search method that uses
classification results of input queries. FAQs aim at covering frequently asked topics and
users usually search topics in FAQs with queries represented by bag-of-words or natural language sentences. However, there is a problem that each question in FAQs is not
usually sufficient enough to cover variety of queries that have the similar meaning but
different surface expressions, such as synonyms, paraphrase and causal relations due
to each topic usually consists of a representative question and its answer. As a result,
users who cannot find their answers in FAQs ask a call center operator. To consider
similarity of meaning among different surface expressions, we use a document classier
that classifies each query into topics of FAQs. A document classifier is trained with not
only FAQs but also corresponding histories of operators for covering variety of queries. However, corresponding histories do not include links to FAQs, we use a method
for generating training data from the corresponding histories with FAQs. To generate
training data correctly, the method takes advantage of a characteristic that many answers in corresponding histories related to FAQs are created by quoting corresponding
FAQs. Our method uses a surface similarity between answers in corresponding histories and the answer part of each topic in FAQs for automatically generating training
data. Experimental results show that our method outperforms an FAQ search based
method using word matching in terms of Mean Reciprocal Rank and Precision@N.
121
An Investigation of Objective Interestingness Measures for Association
Rule Mining
Ratchasak Somyanonthanakul1 and Thanaruk Theeramunkong2
1Department of Medical Informatics, College of Information and Communication
Technology, Rangsit University, Thailand
[email protected]
2Sirindhorn International Institute of Technology, Thammasat University, Thailand
[email protected]
Abstract. While a large number of objective interestingness measures have been proposed
to describe an association pattern which encodes meaningful relationship among attributes
in a dataset, their characteristics and interrelations are not well explored. In this work,
we investigate static and dynamic characteristics of 21 commonly used interestingness
measures in order to understand their common and distinct properties. Four systematical
methods investigated are (1) trend analysis, (2) fixed-total variable-portion analysis, (3)
fixed-total fixed-portion-combination analysis, and (4) imbalance and extreme scenario
analysis. A correlation analysis has been made to find interrelation patterns of the measures.
Keywords: Association rules, interestingness, measure analysis.
122
An Orientation Histogram based Approach for Fall Detection using
Wearable Sensors
Nguyen Ngoc Diep, Cuong Pham, and Tu Minh Phuong
Computer Science Department,
Posts and Telecommunications Institute of Technology, Vietnam
Machine Learning & Applications Lab,
{diepnguyenngoc,cuongpv,phuongtm}@ptit.edu.vn
Abstract. Histogram features are extracted by calculating the distribution of orientations of small fragments or quanta of sliding windows on the sensors continuously acceleration data stream. Bins of the histogram is automatically computed based
on clusters of similar orientations of quanta, making it less sensitive to parameters
used in selection of bins than a heuristic approach. We also present a ner representation of the sliding window by applying the above extraction method to extract local feature vectors of small data segments instead of calculating features from the
whole sliding window. Extracted features are used with support vector machines
trained to classify frames of data streams into containing falls or non-falls. We
evaluated the proposed method on three public datasets with acceleration data including falls and other activities of daily living. On all three datasets, performance
of the proposed method is substantially higher than two other fall detection methods.
Keywords: Fall detection, wearable sensors, SVM, feature extraction.
123
BDSCyto: an Automated Approach for Identifying Cytokines Based on
Best Dimension Searching
Quan Zou1, Shixiang Wan1, Bing Han2, Zhihui Zhan3
1School of Computer Science and Technology, Tianjin University, Tianjin, China
[email protected]
2School of Electronic Engineering, Xidian University, Xi’an, China
[email protected]
3School of Computer Science and Engineering, South China University of Technology,
Guangzhou, China, [email protected]
Abstract. We proposed an automated method for distinguishing cytokines from other proteins according to their primary sequences. Two strategies were employed to
extract features from protein sequences. The first one is a single method, which includes autocorrelation and pseudo amino acid composition extracted feature methods
based on composition and physical–chemical properties of proteins; while the second one is an optimal dimension searching method. Moreover, we developed BDSCyto as a web server to help researchers in classifying protein sequences efficiently
and accurately. BDSCyto reduces the processing time and offers high accuracy by a
series of efficient methods and multithreading technology based on Spark for largescale data. Currently, numerous methods exceed 90% accuracy in cytokine protein
prediction, which is better than the existing single methods. BDSCyto is an opensource project and can be freely accessed by the public at http://bdscyto.sinaapp.com/.
124
Capability-aware Trust Evaluation Model in Multi-agent Systems
Tung Doan Nguyen, Quan Bai, and Weihua Li
School of Engineering, Computer and Mathematical Sciences,
Auckland University of Technology, New Zealand,
{tung.nguyen, quan.bai, weihua.li}@aut.ac.nz
Abstract. Modeling trust in a real time of dynamic multi-agent systems is important
but challenging, particularly when agents frequently join and leave, and the structure of the society may often change. With the increasing complexity of services,
some simplied assumptions, e.g., unlimited processing capability, adopted by several trust models have shown their limitations which restrict the application of trust
model in real-world situations. This paper attempts to relax the unlimited processing
capability assumption of agents by introducing a capability-aware trust evaluation
with temporal factor using hidden Markov model. The experimental results show
that the approach not only can improve the accuracy of trust computation but also
benet the trust-aware decision making for both individual and agent group context.
Keywords: multi-agent system, trust, composite services, capability-aware
125
Combining RDR-based Machine Learning Approach and Human Expert
Knowledge for Phishing Prediction
Hyunsuk Chung, Renjie Chen, Soyeon Caren Han, and Byeong Ho Kang
School of Engineering and ICT, Tasmania 7005, Australia
{David.Chung, renjiec, Soyeon.Han, Byeong.Kang}@utas.edu.au
Abstract. Detecting phishing websites has been noted as complex and dynamic problem area because of the subjective considerations and ambiguities of detection mechanism. We propose a novel approach that uses Ripple-down Rule (RDR) to acquire
knowledge from human experts with the modified RDR model-generating algorithm (Induct RDR), which applies machine-learning approach. The modified algorithm considers two different data types (numeric and nominal) and also applies information theory from decision tree learning algorithms.Our experimental results
showed the proposing approach can help to deduct the cost of solving over-generalization and over-fitting problems of machine learning approach. Three models were
included in comparison: RDR with machine learning and human knowledge, RDR
machine learning only and J48 machine learning only. The result shows the improvements in prediction accuracy of the knowledge acquired by machine learning.
Keywords: phishing prediction, RDR; knowledge-based system; machine learning;
decision tree
126
Combining Swarm with Gradient Search for Maximum Margin
Matrix Factorization
Salman K.H1, Arun K Pujari12, Vikas Kumar1 and Sowmini Devi V.1
1School of Computer and Information Sciences, University of Hyderabad, India
2Central University of Rajasthan, India
[email protected], [email protected], [email protected],
[email protected]
Abstract. Maximum Margin Matrix Factorization is one of the very popular techniques
of collaborative ltering. The discrete valued rating matrix with a small portion of known
ratings is factorized into two latent factors and the unknown ratings are estimated by the
resulting product of the factors. The factorization is achieved by optimizing a loss function and the optimization is carried out by gradient descent or its variants. It is observed
that any of these algorithms yields near-global optimizing point irrespective of the initial
seed point. In this paper, we propose to combine swarm-like search with gradient descent
search. Our algorithm starts from multiple initial points and uses gradient information
and swarm-search as the search progresses. We show that by this process we get an efficient search scheme to get near optimal point for maximum margin matrix factorization.
Computing Probabilistic Assumption-based Argumentation
Nguyen Duy Hung
Sirindhorn International Institute of Technology, Thailand
[email protected]
Abstract. We develop inference procedures for a recently proposed model of probabilistic argumentation called PABA, taking advantages of well-established dialectical proof procedures for Assumption-based Argumentation and Bayesian Network algorithms. We establish the soundness and termination of our
inference procedures for a general class of PABA frameworks. We also discuss how
to translate other models of probabilistic argumentation into this class of PABA
frameworks so that our inference procedures can be used for these models as well.
Keywords: Probabilistic Argumentation, InferenceProcedures,Bayesian Networks
127
Detecting Critical Links in Complex Network to Maintain Information
Flow/Reachability
Kazumi Saito1, Masahiro Kimura2, Kouzou Ohara3, and Hiroshi Motoda45
1School of Administration and Informatics, University of Shizuoka
[email protected]
2Department of Electronics and Informatics, Ryukoku University
[email protected]
3Department of Integrated Information Technology, Aoyama Gakuin University
[email protected]
4Institute of Scientific and Industrial Research, Osaka University
[email protected]
5School of Computing and Information Systems, University of Tasmania
Abstract. We address the problem of efficiently detecting critical links in a large network. Critical links are such links that their deletion exerts substantial effects on the network performance. Here in this paper, we define the performance as being the average
node reachability. This problem is computationally very expensive because the number
of links is an order of magnitude larger even for a sparse network. We tackle this problem by using bottom-k sketch algorithm and further by employing two new acceleration
techniques: marginal-link updating (MLU) and redundant-link skipping (RLS). We tested the effectiveness of the proposed method using two real-world large networks and two
synthetic large networks and showed that the new method can compute the performance
degradation by link removal about an order of magnitude faster than the baseline method
in which bottom-k sketch algorithm is applied directly. Further, we confirmed that the
measures easily composed by well known existing centralities, e.g. in/out-degree, betweenness, PageRank, authority/hub, are not able to detect critical links. Those links detected by these measures do not reduce the average reachability at all, i.e. not critical at all.
Keywords: Social networks, Link deletion, Critical links, Node reachability
128
Distributed B-SDLM: Accelerating the Training Convergence of Deep
Neural Networks through Parallelism
Shan Sung Liew1, Mohamed Khalil-Hani1, and Rabia Bakhteri2
1VeCAD Research Laboratory Faculty of Electrical Engineering
Universiti Teknologi Malaysia 81310 Skudai, Johor Malaysia
2Machine Learning Developer Group Sightline Innovation
#202, 435 Ellice Ave Winnipeg MB R3B 1Y6 Canada
[email protected], [email protected], [email protected]
Abstract. This paper proposes an efficient asynchronous stochastic second order learning algorithm for distributed learning of neural networks (NNs). The proposed algorithm, named distributed bounded stochastic diagonal Levenberg-Marquardt (distributed
B-SDLM), is based on the BSDLM algorithm that converges fast and requires only minimal computational overhead than the stochastic gradient descent (SGD) method. The
proposed algorithm is implemented based on the parameter server thread model in the
MPICH implementation. Experiments on the MNIST dataset have shown that training
using the distributed B-SDLM on a 16-core CPU cluster allows the convolutional neural
network (CNN) model to reach the convergence state very fast, with speedups of 6:03 and
12:28 to reach 0:01 training and 0:08 testing loss values, respectively. This also results in
significantly less time taken to reach a certain classification accuracy (5:67 and 8:72 faster
to reach 99% training and 98% testing accuracies on the MNIST dataset, respectively).
Keywords: Deep learning, distributed machine learning, stochastic diagonal Levenberg-Marquardt, convolutional neural network.
129
Early Detection of Osteoarthritis Using Local Binary Patterns: A Study
Directed at Human Joint Imagery
Kwankamon Dittakan1 and Frans Coenen2
1Faculty of Technology and Environment,
Prince of Songkla University, Phuket Campus,
80 Moo 1, Vichit-Songkram Road, Kathu, Phuket, Thailand
2Department of Computer Science,
University of Liverpool, Liverpool, L69 3BX, United Kingdom
{[email protected],[email protected]}
Abstract. Osteoarthritis (OA) is a chronic health condition that causes severe joint
pain and stiffness; it is a major cause of disability in older people. The risk of OA
increases from age 45 and older. Early diagnosis is typically made using X-ray imagery. In this paper an automated mechanism for OA screening is proposed. The
fundamental idea is to generate a classifier that is able to distinguish between OA or
non-OA images. The challenge is how bast to translate an X-ray image into a form
that serves to both captures key information while remaining compatible with the
classification process. It is suggested that image texture is the most desirable feature to be considered. The process is filly described and evaluated. The data used
for the evaluation was obtained from the right Tibia of 50 female subjects. Excellent results were obtained, recorded AUC values of 1:0. Keywords: Data Mining,
Image Classification, Medical Image Analysis and Mining, Osteoarthritis Screening
130
Evaluation of Deposit-based Road Pricing Scheme by
Agent-based Simulator
Ryo KANAMORI1,*, Toshiyuki YAMAMOTO2, Takayuki MORIKAWA1
1Nagoya University, Nagoya, Japan
{kanamori.ryo, morikawa}@nagoya-u.jp
2Nagoya University, Nagoya, Japan
[email protected]
Abstract. Parking Deposit System (PDS) is proposed to improve public acceptance of
road pricing (RP), which can reduce effectively traffic congestion and air pollution. This
study examines the characteristics of PDS and conventional cordon-based RP in terms of
efficiency and equity. In order to evaluate income regressive effects as equity, we develop a
multi-agent based simulator which can consider user’s mode and route choice behaviors
for several income classes. The results of empirical analysis at the Nagoya Metropolitan
Area suggest the followings in comparison with conventional RP: 1) PDS gives sufficient
environmental improvement by reducing car use, although the effect of decreases by refund, and 2) PDS is more equitable because it is a kind of revenue allocation scheme.
Keywords: Traffic problem, Road pricing, Agent based simulation
Exploring Multi-Action Relationship in Reinforcement Learning
Han Wang and Yang Yu*
National Key Laboratory for Novel Software Technology,
Nanjing University, Nanjing 210023, China
{wangh,yuy}@lamda.nju.edu.cn
Abstract. In many real-world reinforcement learning problems, an agent needs to control multiple actions simultaneously. To learn under this circumstance, previously, each
action was commonly treated independently with other. However, these multiple actions
are rarely independent in applications, and it could be helpful to accelerate the learning if
the underlying relationship among the actions is utilized. This paper explores multi-action relationship in reinforcement learning. We propose to learn the multi-action relationship by enforcing a regularization term capturing the relationship. We incorporate
the regularization term into the least-square policy-iteration and the temporal-difference
methods, which result efficiently solvable convex learning objectives. The proposed
methods are validated empirically in several domains. Experiment results show that incorporating multi-action relationship can effectively improve the learning performance.
131
Face Verication Algorithm with Exploiting
Feature Distribution
Zhi Qu, Xuan Li, Yong Dou*, and Ke Yang.
Key Laboratory of Parallel Distribution
SanYi avenue, Changsha, Hunan province, China
[email protected]
Abstract. Deep feature is widely applied in many elds such as image retrieval, image classication, face verication, etc. All the post-processing methods using deep feature make some assumptions about feature distribution. However, in most situations,
features do not follow the hypothesised distribution approximately. In this paper, we
focus on face-verication applications which also suer from these problems. We propose an up-sample method called IUSM to alleviate the problems caused by biased
samples. Additionally, by analyzing the Joint Bayesian model theoretically and practically, we propose a feature fusion method called LFF which utilizes the distribution properties of Joint Bayesian. Based on IUSM, the face verication accuracy of
biased data is improved by 6% while generalization ability of convolution network
is not crippled. On the widely used Labeled Face in the Wild(LFW) dataset, LFF
method can slightly improve the accuracy 0.15% while the baseline accuracy is more
than 97.51%. We also argue that LFF can improve each deep face verication algorithm which uses Joint Bayesian model due to LFF’s linear combination of features.
132
Facial Age Estimation by Total Ordering Preserving Projection
Xiao-Dong Wang and Zhi-Hua Zhou
National Key Laboratory for Novel Software Technology
{wangxd,zhouzh}@lamda.nju.edu.cn
Abstract. Facial age estimation is one of the unsolved challenging issues in automatic face perception. Previous studies usually formulated it as a classication problem,
where each age is regarded as a class, or a regression problem where the age is regarded as a variable spanning in a real-valued interval. In this paper, we propose to
formulate this task as an ordinal regression problem. On one hand, the new formulation emphasizes the fact that the age estimation problem is inherently a classication
problem (ordinal regression is a special kind of classication task); on the other hand,
the new formulation allows to take into account the order information between dierent ages, which has been ignored by previous classication formulation. We develop
the TOPP (Total Ordering Preserving Projection) approach, by identifying the low-dimensional subspace which preserves the ordinal relations to the best, and experiments
show that TOPP signicantly outperforms state-of-the-art age estimation methods.
Keywords: ordering preserving projection, ordinal regression, facial ageestimation,
face perception, machine learning
133
Fast Training of a Graph Boosting for Large-Scale Text Classication
Hiyori Yoshikawa and Tomoya Iwakura
Fujitsu Laboratories Ltd., Kawasaki, Japan
{y.hiyori,iwakura.tomoya}@jp.fujitsu.com
Abstract. This paper proposes a fast training method for graph classication based on
a boosting algorithm and its application to sentimental analysis with input texts represented by graphs. Graph format is very suitable for representing texts structured with
Natural Language Processing techniques such as morphological analysis, Named Entity
Recognition, and parsing. A number of classication methods which represent texts as
graphs have been proposed so far. However, many of them limit candidate features in
advance because of quite large size of feature space. Instead of limiting search space in
advance, we propose two approximation methods for learning of graph-based rules in
a boosting. Experimental results on a sentimental analysis dataset show that our method contributes to improved training speed. In addition, the graph representation-based
classication method exploits rich structural information of texts, which is impossible
to be detected when using other simpler input formats, and shows higher accuracy.
Keywords: text classication feature engineering graph boosting
Faster Convergence to Cooperative Policy by Autonomous Detection of
Interference States in Multiagent Reinforcement Learning
Sachiyo Arai and Haichi Xu
Faculty of Engineering, Chiba University,
1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, JAPAN
Abstract. In this paper, we propose a method for ameliorating the state-space explosion
that can occur in the context of multiagent reinforcement learning. In our method, an agent
considers other agents’ states only when they interfere with each other in attaining their
goals. Our idea is that the initial state-space of each agent does not include information about
other spaces. Agents then automatically expand their state-space if they detect interference states. We adopt the information theory measure of entropy to detect the interference
states for which agents should consider the state information of other agents. We demonstrate the advantage of our method with respect to the efficiency of global convergence.
Keywords: Multiagent system, Reinforcement learning, Conflict resolution
134
Generalized Extreme Value Filter to Remove Mixed
Gaussian-Impulse Noise
Sakon Chankhachon1, Sathit Intajag2,*
Artificial Intelligence Research Laboratory,
Department of Computer Science, Faculty of Science,
Prince of Songkla University, Thailand.
1([email protected]), 2([email protected])
*Corresponding author
Abstract. Noise removal in image restoration is an important technique of image
processing. In this paper, a new efficient approach is proposed for removing the mixed
Gaussian-impulse noise in a color image. The proposed method utilizes the concept of
local rank ordered absolute distances to measure similarity between a processing pixel
in the small window and their neighborhood pixels in the processing block. The generalized extreme value distribution was employed to estimate weighted averages of the
pixels in the processing block for filtering the mixed Gaussian-impulse noise. From the
experimental results, our filter has yielded the better results in suppressing high density
levels of the mixed noise in the color images than the state-of-the-art denoising methods.
Keywords: generalized extreme value; mixed Gaussian-impulse noise; local rank
ordered absolute distances.
Generating Covering Arrays with pseudo-Boolean Constraint Solving and
Balancing Heuristic*
Hai Liu2, Feifei Ma1, Jian Zhang1
1State Key Laboratory of Computer Science
Institute of Software, Chinese Academy of Sciences
Email: fma,[email protected]
2Beijing Information Science and Technology University
Abstract. Covering arrays (CAs) are interesting objects in combinatorics and they also
play an important role in software testing. It is a challenging task to generate small
CAs automatically and efficiently. In this paper, we propose a new approach which
generates a CA column by column. A kind of balancing heuristic is adopted to guide
the searching procedure. At each step (column extension), some pseudo Boolean constraints are generated and solved by a PBO solver. A prototype tool is implemented, which turns out to be able to nd smaller CAs than other tools, for some cases.
135
Grouped Text Clustering Using Non-Parametric
Gaussian Mixture Experts
Yong Tian, Yu Rong, Yuan Yao, Weidong Liu, and Jiaxing Song
Department of Computer Science and Technology,
Tsinghua University, Beijing, China
[email protected], [email protected]
Abstract. Text clustering has many applications in various areas. Before being clustered, texts often have already been grouped or partially grouped in practise. Texts
from the same group are related to each other and concentrate on a few topics. The
group information turns out to be valuable for text clustering. In this paper, we propose a model called Non-parametric Gaussian Mixture Experts to get better clustering result through utilizing group information. After converting texts to vectors by
semantic embedding, our model can automatically infer proper cluster number for
every group and the whole corpus. We develop an online variational inference algorithm which is scalable and can handle incremental datasets. Our algorithm is tested on
various text datasets. The results demonstrate our model has signicantly better performance in cluster quality than some other classical and recent text clustering methods.
Hybrid Temporal-Dierence Algorithm using Sliding Mode Control and
Sigmoid Function
Ke Xu and Fengge Wu
Science and Technology on Integrated Information System Laboratory
Institute of Software Chinese Academy of Sciences
{xuke13,fengge}@iscas.ac.cn
Abstract. Gradient temporal-dierent algorithms such as GTD2 and TDC have improved
the accuracy of the algorithm to a new level. Unfortunately, these algorithms converge
much slower than conventional temporal-dierent algorithms. In this paper, we present a
approach based on sliding mode control to speed up the GTD2 algorithm, and then use sigmoid function to reduce algorithm’s jitter. Our experiments on random walk show that our
algorithm converges as fast as conventional temporal-dierent algorithms and as accurate as
GTD2 algorithm at the same time. This is an important property for online-learning tasks.
Keywords: reinforcement learning, temporal-dierence algorithm, sliding mode control, sigmoid function
136
Incorporating an Implicit and Explicit Similarity Network for User-level
Sentiment Classication of Microblogging
Yongyos Kaewpitakkun and Kiyoaki Shirai
Japan Advanced Institute of Science and Technology
{s1320203,kshirai}@jaist.ac.jp
Abstract. In Twitter, the sentiments of individual tweets are dicult to classify, but
the overall opinion of a user can be determined by considering their related tweets
and their social relations. It would be better to consider not only the textual information in the tweets, but also the relationships between the users. Previous approaches that incorporate network information into the classier have mainly focussed
on \a link” dened by the explicitly connected network, such as, follow, mention, or
retweet. However, the presence of explicit link structures in some social networks
is limited. In this paper, we propose a framework that takes into consideration the \
implicit connections” between users. An implicit connection refers to the relations
of users who share similar topics of interest, as extracted from their historical tweet
corpus, which contains much data for analysis. The results of experiments show that
our method is effective and improves the performance compared to the baselines.
Keywords: Sentiment Analysis, Factor-graph Model, Topic Modeling, Machine
Learning, Microblogging
137
Information Retrieval from Unstructured Arabic Legal Data
Imen Bouaziz Mezghanni and Faiez Gargouri
MIRACL Laboratory
ISIM Sfax, TUNISIA
{[email protected],
[email protected]}
http://www.miracl.rnu.tn/
Abstract. Given the steady increase of published and stored information in the form of
Arabic unstructured texts, current Information Retrieval (IR) systems must be able to
suit the nature and requirements of this language for an accurate and efficient search. This
paper sheds light on the challenges in Arabic IR (AIR) and proposes an approach for enhancing the process of AIR based on transforming these texts into structured documents
in XML format through a document ontology as well as a set of linguistic grammars. The
IR system hence is done on the XML documents. The aim of such system is to incorporate
the knowledge on the document structure and on specic content elements in computing
the relevance of an information element. A query expansion module mainly based on domain ontology as well as user prole is proposed for the enhancement of the search results.
Keywords: Information retrieval, Arabic information retrieval, Unstructured data,
Structured data
138
Instance Selection Method for Improving Graph-based
Semi-Supervised Learning*
Hai Wang, Shao-Bo Wang, and Yu-Feng Li
National Key Laboratory for Novel Software Technology, Nanjing University,
Nanjing, 210023, China
{wanghai,wangsb,liyf}@lamda.nju.edu.cn
Abstract. Graph-based semi-supervised learning (GSSL) is one of the most important semi-supervised learning (SSL) paradigms. Though GSSL methods are helpful in
many situations, they may hurt performance when using unlabeled data. In this paper,
we propose a new GSSL method GsslIs based on instance selection in order to reduce the chances of performance degeneration. Our basic idea is that given a set of
unlabeled instances, it is not the best to exploit all the unlabeled instances; instead,
we should exploit the unlabeled instances which are highly possible to help improve
the performance, while do not take the ones with high risk into account. Experiments on a board range of data sets show that the chance of performance degeneration of our proposal is much smaller than that of many state-of-the-art GSSL methods.
Keywords: graph-based semi-supervised learning performance degeneration instance
selection
139
L1-Regularized Continuous Conditional Random Fields
Xishun Wang1, Fenghui Ren1, Chen Liu2, and Minjie Zhang1
1School of Computing and Information Technology,
2School of Science, RMIT University, Australia
[email protected],
Abstract. Continuous Conditional Random Fields (CCRF) has been widely applied to
various research domains as an efficient approach for structural regression. In previous
studies, the weights of CCRF are constrained to be positive from a theoretical perspective. This paper extends the definition domains of weights of CCRF and thus introduces
L1 norm to regularize CCRF, which enables CCRF to perform feature selection. We
provide a plausible learning method for L1-Regularized CCRF (L1- CCRF) and verify
its effectiveness. Moreover, we demonstrate that the proposed L1-CCRF performs well
in selecting key features related to the various customers’ power usages in Smart Grid.
Keywords: Continuous Conditional Random Fields; Regularization; Feature selection
140
Large Margin Coupled Mapping for Low Resolution Face Recognition
Jiaqi Zhang1, Zhenhua Guo134, Xiu Li1, and Youbin Chen2
1Graduate School at Shenzhen, Tsinghua University, Shenzhen, China
2Huazhong University of Science and Technology, China
3Key Laboratory of Measurement and Controlof Complex Systems of Engineering,
Ministry of Education, Southeast University, China
4Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information, Ministry of Education, Nanjing University of Science and Technology, China
Abstract. Traditional face recognition algorithms can achieve significant performance under well-controlled environments. However, these algorithms perform
poorly when the resolution of the face images varies. A two-step framework is proposed to solve the resolution problem through adopting super-resolution (SR) and
performing face recognition on the super-resolved face images. However, such method usually has poor performance on recognition tasks as SR focuses more on visual enhancement, rather than classification accuracy. Recently, Coupled Mapping
(CM) has been introduced into face recognition framework across different resolutions, which learns a common feature subspace for both high-resolution (HR) and
low-resolution (LR) face images. In this paper, inspired by maximum margin projection, we propose Large Margin Coupled Mapping (LMCM) algorithm, which learns
projections to maximize the margin between distance of between-class subjects and
distance of within-class ones in the common space. Experiments on public FERET and SCface databases demonstrate that LMCM is effective for low-resolution face recognition.
Keywords: Coupled Mapping, Low-Resolution Face Recognition, Large Margin
Coupled Mapping, FERET, SCface
141
Learning from Numerous Untailored Summaries
Yuta Kikuchi, Akihiko Watanabe, Sasano Ryohei,
Hiroya Takamura, and Manabu Okumura
{kikuchi,watanabe,sasano,takamura,oku}@lr.pi.titech.ac.jp
Abstract. We present an attempt to use a large amount of summaries contained in
the New York Times Annotated Corpus (NYTAC). We introduce five methods inspired by domain adaptation techniques in other research areas to train our supervised summarization system and evaluate them on three test sets. Among the
five methods, the one that is trained on the NYTAC followed by fine-tuning on
the target data (i.e. the three test sets; DUC2002, RSTDTBlong and RSTDTBshort) performs the best for all the test sets. We also propose an instance selection
method according to the faithfulness of the extractive oracle summary to the reference summary and empirically show that it improves summarization performance.
Learning of Evaluation Functions to Realize Playing Styles in Shogi
Shotaro Omori* and Tomoyuki Kaneko
Graduate School of Arts and Sciences, the University of Tokyo,
{omori,kaneko}@graco.c.u-tokyo.ac.jp
Abstract. This paper presents a method to give a computer player an intended playing style by the machine learning of an evaluation function. Recent improvements in
machine learning techniques have realized the automated tuning of the feature weight
vector of an evaluation function. To make a strong player, as many moves as possible of
strong players’ game records are needed, though the number of available game records
decreases when we focus on a specic playing style. To pursue both goals of playing
style and playing strength, we present three steps of learning: classifying moves with
respect to playing styles, training the weight vector of an evaluation function by using
the whole set of game records to maximize its playing strength, and modifying the
weight vector carefully so as to improve agreement with the moves of the intended
playing style. We applied our method to realize players of defense or attack-oriented style in shogi and tested the players by self-play against the original version. The
results conrmed that the presented method successfully adjusted evaluation functions
in that the frequency of defensive moves is signicantly increased or decreased in accordance with the game records used while keeping the winning ratio at almost 50%.
142
Learning Sentimental Weights of Mixed-gram Terms for
Classication and Visualization
Tszhang Guo12 Bowen Li1, Zihao Fu3, Tao Wan14, and Zengchang Qin1
1Intelligent Computing and Machine Learning Lab, School of ASEE,
Beihang University, Beijing, 100191, China
2Department of Automation, Tsinghua University, Beijing, China
3Alibaba Group, Beijing, 100022, China
4School of Biological Science and Medical Engineering, Beihang University
{tao.wan.wan,libowen.ne,zengchang.qin}@gmail.com
Abstract. Sentimental analysis is an important topic in natural language processing and
opinion mining. Many previous studies have reported to judge whether a term is with
emotion or not. However, little work has been done in measuring degrees of sentiment
for these terms. For example, the word excellent has stronger positive sentiment than the
word good and okay. In this paper, we investigate how to model this intricate sentimental
difference by assigning sentimental weights. A simple and effective model is proposed
based on logistic regression to extract emotional terms associated with sentiment weights.
Weighted terms can be used in sentiment classification and visualization by drawing
emotional clouds of texts. The new model is tested using uni-gram, bi-gram and mixedgram language models on two benchmark datasets. The empirical results show that the
new model is highly efficient with comparable accuracy to other sentiment classifiers.
Learning with Additional Distributions
Sanparith Marukatat
IMG lab, NECTEC
112 Thailand Science Park, Pathumthani, Thailand
[email protected]
Abstract. This paper studies the problem of learning with distributions.
In this work, we do not focus on the distribution that represents each data point. Instead, we consider the distribution that is an additional information around each data
point. The proposed method yields a new kernel that is similar to an existing one. The
main dierence is that our kernel requires an integration in the kernel space. Theoretically, the proposed method yields a better generalization compared to normal SVM.
Keywords: SVM, SMM, Kernel for distributions
143
Local Search with Noisy Strategy for Minimum Vertex Cover in
Massive Graphs
Zongjie Ma1*, Yi Fan1, Kaile Su1, Chengqian Li2, and Abdul Sattar1
1Institute for Integrated and Intelligent Systems, Grith University, Brisbane, Australia
2Department of Computer Science, Sun Yat-sen University, China
[email protected]
Abstract. Finding minimum vertex covers (MinVC) for simple undirected graphs
is a well-known NP-hard problem. In the literature there have been many heuristics for obtaining good vertex covers. However, most of them focus on solving this
problem in relatively small graphs. Recently, a local search solver called FastVC is
designed to solve the MinVC problem on real-world massive graphs. Since the traditional best- picking heuristic was believed to be of high complexity, FastVC replaces it with an approximate best-picking strategy. However, since best-picking has
been proved to be powerful for a wide range of problems, abandoning it may be a
great sacrice. In this paper we have developed a local search MinVC solver which
utilizes best-picking with noise to remove vertices. Experiments conducted on a
broad range of real-world massive graphs show that our proposed method finds better vertex covers than state-of-the-art local search algorithms on many graphs.
Keywords: Minimum Vertex Cover, Heuristic Search, Massive Graphs, Combinatorial Optimization, Social Networks
144
Maximum Margin Tree Error Correcting Output Codes
Fa Zheng12, Hui Xue12*, Xiaohong Chen3, and Yunyun Wang4
1School of Computer Science and Engineering, Southeast University, Nanjing,
210096, P.R. China
2Key Laboratory of Computer Network and Information Integration (Southeast
University), Ministry of Education, P.R. China
3College of Science, Nanjing University of Aeronautics and Astronautics, Nanjing,
210016, P.R. China
4Department of Computer Science and Engineering, Nanjing University of Posts and
Telecommunications, Nanjing, 210046, P.R. China
{faaronzheng,hxue}@seu.edu.cn,[email protected],[email protected]
Abstract. Encoding is one of the most important steps in Error Correcting Output
Codes (ECOCs). Traditional encoding strategies are usually data-independent. Recently, some tree-form encoding algorithms are proposed which rstly utilize mutual
information to estimate inter-class separability in order to create a hierarchical partition of the tree from top to down and then obtain a coding matrix. But such criterion is usually computed by a non-parametric method which would generally require
vast samples and is more likely to lead to unstable results. In this paper, we present
a novel encoding algorithm which uses the maximum margins between classes as
the criterion and constructs a bottom-up binary tree based on the maximum margin. As a result, the corresponding coding matrix is more stable and discriminative for the following classication. Experimental results have shown that our algorithm performs much better than some state-of-the-art coding algorithms in ECOC.
Keywords: Multi-class Classication, Maximum Margin Tree, Error Correcting Output
Codes.
145
Modeling of Travel Behavior Processes from Social Media
Yuki Yamagishi, Kazumi Saito, and Tetsuo Ikeda
School of Management and Information University of Shizuoka
[email protected], {k-saito, t-ikeda}@u-shizuoka-ken.ac.jp
Abstract. We attempt stochastic modeling of travel behavior processes from the observed data. To this end, based on the Le´vy flight behavior process combined with
the popularity of each point of interest, we first propose a probability model and efficient method that estimates the model parameters from the observed user behavior
data. Then, we propose two methods for POI ranking by using the probability obtained from our proposed model. In our experiments using user behavior data constructed from a review site dataset, we report our experimental results on parameter
estimation and examine the properties of POI ranking methods in comparison to a
naive popularity ranking method. As our experimental results, we show that our parameter estimation results are intuitively interpretable, and as a favorable property,
our ranking methods naturally give high ranks to POIs located in attractive regions.
Keywords: travel behavior processes, social media, probability model
146
Motion Primitive Forests for Human Activity Recognition using Wearable
Sensors
Nguyen Ngoc Diep, Cuong Pham, and Tu Minh Phuong
Computer Science Department,
Machine Learning & Applications Lab,
{diepnguyenngoc,cuongpv,phuongtm}@ptit.edu.vn
Abstract. Human activity recognition is important in many applications such as fitness
logging, pervasive healthcare, near-emergency warn ing, and social networking. Using
body-worn sensors, these applications detect activities of the users to understand the
context and provide them appropriate assistance. For accurate recognition, it is crucial
to design appropriate feature representation of sensor data. In this paper, we propose a
new type of motion features: motion primitive forests, which are randomized ensembles
of decision trees that act on original l cal features by clustering them to form motion
primitives (or words). The bags of these features, which accumulate histograms of the
resulting motion primitives over each data frame, are then used to build activity models.
We experimentally validated the effectiveness of the proposed method on accelerometer
data on three benchmark datasets. On all three datasets, the proposed motion primitive
orests provided substantially higher accuracy than existing state-of-the-art methods, and
were much faster in both training and prediction, compared with k-means feature learning.In addition, the method showed stable results over different types of original local
features, indicating the ability of random forests in selecting relevant local features.
Keywords: human activity recognition, wearable sensors, motion primitive forests,
random forests, bag of features.
147
Multi-Level Occupancy Grids for Efficient
Representation of 3D Indoor Environments
Yu Tian, Wanrong Huang, Yanzhen Wang, Xiaodong Yi,
Zhiyuan Wang, and Xuejun Yang
State Key Laboratory of High Performance Computing (HPCL), College of Computer
137 Yanwachi Street, Changsha, Hunan, P. R. China, 410073
[email protected]
Abstract. Mapping 3D environments is a fundamental yet challenging problem for
mobile robot applications. Although 3D sensory data can be effciently obtained using low-cost commercial RGB-D cameras, direct extension of the widely-adopted
occ pancy grids to 3D environments would cause problems, such as large storage co
sumption and intensive computation cost. In this paper, we propose to use a stack of
2D occupancy grids, each of which corresponds to a horizontal slice of the 3D environment at a specic height, as an efficient representation of 3D environments for indoor applications. Moreover, an existing algorithm based on Rao-Blackwellized Particle Filters (RBPF) is modied accordingly to perform simultaneous localization and
mapping (SLAM) using the proposed multi-level occupancy grids (M-LOG), the entire codes of which have been made open source at https://github.com/AngelTianYu/
micros_mlog. Experimental results from both simulation and real-world tests validate the effectiveness of the proposed approach in indoor environments. Computational cost of the approach scales linearly with the number of 2D map slices, making it the user’s choice the trade-off between vertical map resolution and efficiency.
Keywords: simultaneous localization and mapping; particle lters; mobile robots;multi-level occupancy grids
148
Multi-view Representative and Informative induced Active Learning
Huaxi Huang, Changqing Zhang*, Qinghua Hu, Pengfei Zhu
School of Computer Science and Technology, Tianjin University, Tianjin, China
{hhx10,zhangchangqing,huqinghua,zhupengfei}@tju.edu.cn
Abstract. Most existing active learning methods often manually label samples and train
models with labeled data in an iterative way. Unfortunately, at the early stage of the experiment, few labeled data are available, hence, selecting the most valuable data points to
label is necessary and important. To this end, we propose a novel method, called Multiview Representative and Informative-induced Active Learning (MRI-AL), which selects
samples of both representativness and informativeness with the help of complementarity
of multiple views. Specically, subspace reconstruction with structure sparsity technique is
employed to ensure the selected samples to be representative, while the global similarity
constraint guarantees the informativeness of the selected samples. The proposed method
is solved efficiently by alternating direction method of multipliers (ADMM). We empirically show that our method outperforms existing early experimental design approaches.
Keywords: Active Learning, Multi-view, Representative, Informative,Subspace
Learning, Structure sparsity
149
Offline Text and Non-text Segmentation for Hand-Drawn Diagrams
Buntita Pravalpruk and Matthew M. Dailey
Computer Science and Information Management
Asian Institute of Technology
Klong Luang, Pathumtani, Thailand
http://www.cs.ait.ac.th/
Abstract. Writing and drawing are basic forms of human communication. Handwritten and hand-drawn documents are often used at initial stages of a project. For
storage and later usage, handwritten documents are often converted into a digital
format with a graphics program. Drawing with a computer in many cases requires
skill and more time than less formal handwritten drawings. Even when people have
experience in computer drawing and are familiar with the application, it takes time.
Automatic conversion of images of hand-drawn diagrams into a digital graphic format file could save time in the design process. One of early critical tasks in handdrawn diagram interpretation is segmentation of the diagram into text and non-text
components. In this paper, we compare two approaches for offline text and non-text
segmentation of contours in an image. We describe the feature extraction and classication processes. Our methods obtain 82-86% accuracy. Future work will explore
the application of these techniques in a complete diagram interpretation system.
Keywords: Image recognition, Hand-drawn diagram, hand-written diagram, text and
non-text classication
On Partial Features in the DLF Family of Description Logics
David Toman and Grant Weddell
Cheriton School of Computer Science
University of Waterloo, Canada
{david,gweddell}@cs.uwaterloo.ca
Abstract. The DLF family of description logics are fragments of first order logic with underlying signatures based on unary predicate symbols, called atomic concepts, and unary function symbols interpreted as total functions, called
features. We show how computational properties relating to a key reasoning service for dialects of this family are preserved when (a) unary function symbols are
now interpreted as partial functions, and when (b) a concept constructor is admitted that can characterize circumstances in which partial functions become total.
150
On the Gradient-Based Sequential Tuning of the Echo State Network
Reservoir Parameters
Sumeth Yuenyong
School of Information Technology, Shinawatra University
99 Moo 10 Bang Toey, Sam Khok District, Pathum Thani 12160, Thailand
Abstract. In this paper, the derivative of the input scaling and spectral radius parameters of Echo State Network reservoir are derived. This was achieved by re-writing
the reservoir state update equation in terms of template matrices whose eigenvalues
can be pre-calculated, so the two parameters appear in the state update equation in
the form of simple multiplication which is dierentiable. After that the paper derives the derivatives and then discusses why direct application of these two derivatives in gradient descent to optimize reservoirs in a sequential manner would be ineffective due to the nature of the error surface and the problem of large eigenvalue
spread on the reservoir state matrix. Finally it is suggested how to apply the derivatives obtained here for joint-optimizing the reservoir and readout at the same time.
Optimization of Road Distribution for Traffic System Based on
Vehicle’s Priority
Wen Gu and Takayuki Ito
Nagoya Institute of Technology,
Gokiso-cho, Showa-ku, Nagoya City, Aichi Pref., Japan
[email protected]
[email protected]
Abstract. Instead of making the traffic system work uently by focusing on each car’s way
to choose their routes, in this paper, we proposed a way to make the vehicles avoid being
involved into the traffic congestion by allocating the roads which are regarded as one kind
of resources to the vehicles. In order to make the road allocation fair, we introduce the
parameter to show each vehicle’s priority. We allocate the roads by regarding it as a linear
programming problem and use linear programming to solve it. The experiment was done
by using simulator SUMO and we testied that our proposal can make the vehicles avoid
getting involved into traffic congestion and veried the usefulness of the vehicle’s priority.
Keywords: Traffic Simulation, Road Allocation, Priority, Linear Programming, Optimization
151
Prediction with Confidence in Item Based Collaborative Filtering
Tadiparthi V. R. Himabindu1, Vineet Padmanabhan1, Arun K. Pujari1, and
Abdul Sattar2
1School of Computer & Information Sciences, University of Hyderabad, India.
[vineet,akp][email protected], himaworld [email protected]
2Grith University, Brisbane, Australia
[email protected]
Abstract. Recommender systems can be viewed as prediction systems where we can
predict the ratings which represent users’ interest in the corresponding item. Typically,
items having the highest predicted ratings will be recommended to the users. But users
do not know how certain these predictions are. Therefore, it is important to associate
a condence measure to the predictions which tells users how certain the system is in
making the predictions. Many dierent approaches have been proposed to estimate
confidence of predictions made by recommender systems. But none of them provide guarantee on the error rate of these predictions. Conformal Prediction is a
framework that produces predictions with a guaranteed error rate. In this paper, we
propose a conformal prediction algorithm with item-based collaborative ltering
as the underlying algorithm which is a simple and widely used algorithm in commercial applications. We propose dierent nonconformity measures and empirically determine the best nonconformity measure.We empirically prove validity and efficiency of proposed algorithm. Experimental results demonstrate that the predictive
performance of conformal prediction algorithm is very close to its underlying algorithm with little uncertainty along with the measures of confidence and credibility.
Keywords: Recommender Systems, Conformal Prediction, Confidence, nonconformity measure
152
Restricted Four-valued Semantics for Answer Set Programming
Chen Chen and Zuoquan Lin*
Department of Information Science
School of Mathematical Sciences
Peking University, Beijing 100871, China
Abstract. In answer set programming, an extended logic program may have
no answer set, or only one trivial answer set. In this paper, we propose a new stable model semantics based on the restricted four-valued logic to overcome both inconsistences and incoherences in answer set programming. Our stable models
coincide with classical answer sets when reasoning on consistent and coherent logic programs, and can be solved by transformation in existing ASP solvers. We also
show the connection between our stable models and the extensions of default logic.
Selecting Training Data for Unsupervised Domain Adaptation in Word
Sense Disambiguation
Kanako Komiya1, Minoru Sasaki1 Hiroyuki Shinnou1, Yoshiyuki Kotani2, and
Manabu Okumura3
1Ibaraki University, 4-12-1 Nakanarusawa, Hitachi-shi, Ibaraki, 316-8511 JAPAN,
fkanako.komiya.nlp, minoru.sasaki.01, [email protected]
2Tokyo University of Agriculture and Thechnology, 2-24-16 Naka-cho, Koganei,
Tokyo, 184-8588 JAPAN, [email protected]
3Tokyo Institute of Technology, 4259 Nagatuta Midori-ku Yokohama 226-8503
JAPAN, [email protected]
Abstract. This paper describes a method of domain adaptation, which involves adapting
a classier developed from source to target data. We automatically select the training data
set that is suitable for the target data from the whole source data of multiple domains.
This is unsupervised domain adaptation for Japanese word sense disambiguation (WSD).
Experiments revealed that the accuracies of WSD improved when we automatically selected the training data set using two criteria, the degree of condence and the leave-oneout (LOO)-bound score, compared with when the classier was trained with all the data.
Keywords: Domain adaptation, word sense disambiguation, data selection
153
Sentiment Analysis for Images on Microblogging by Integrating Textual
Information with Multiple Kernel Learning
Junxin Tan1, Mengting Xu1, Lin Shang1, Xiuyi Jia2
1State Key Laboratory for Novel Software Technology,
Nanjing University, Nanjing, China, 210023
[email protected] [email protected] [email protected]
2School of Computer Science and Engineering, Nanjing University of Science and
Technology, Nanjing, China, 210094
[email protected]
Abstract. Image is one of the most important means to express users’ emotions on
microblogging, like Sina Weibo. More and more people post only images on it, due
to the fast and convenient nature of image. Tak- ing a post only using images on microblogging has been a new tendency. Most existing studies about sentiment analysis on microblogging focus on the text, or integrate image as an auxiliary information
into text, so they are not applicable in this scenario. Although a few methods related
to sentiment analysis for image have been proposed, most of them either ignore the
semantic gap between low-level visual features and higher-level image sentiments, or
require a lot of textual information in the phases of both training and inference. This
paper proposes a new sentiment analysis method based on Simple Multiple Kernel
Learning (SimpleMKL). Specically, textual information as a sort of sufficiently emotional source data, we can use it to promote the ability via SimpleMKL to classify
images. And once we get the image classier, none of texts are needed when predicting other unlabelled images. Experimental results show that our proposed method can improve the performance signicantly on data we crawled and labelled from
Sina Weibo. We find that our method not only outperforms some common methods,
like SVM, Naive Bayes, KNN, Random Forest, Adaboost, etc., using the image features of colour, hog, texture, but also outperforms some state-of-the-art methods.
Keywords: Sentiment analysis, Microblogging, Image sentiment, Multiple Kernel
Learning.
154
Set to Set Visual Tracking
Wencheng Zhu, Pengfei Zhu*, Qinghua Hu, and Changqing Zhang
School of Computer Science and Technology, Tianjin University, Tianjin, China
{zhu1992719,zhupengfei,huqinghua,zhangchangqing}@tju.edu.cn
Abstract. Sparse representation has been widely used in visual tracking and achieves superior tracking results. However, most sparse representation models represent the target
candidate as a linear combination of target templates and need to solve a sparse optimization problem. In this paper, we propose a novel set to set visual tracking (SSVT) method. Under the particle lter framework, we consider both the target candidates and target
templates as image sets, and model them as convex hulls. Then the distance between two
image sets is minimized and the tracking result is the target candidate with the maximum
coefficient. As the target candidates are modeled as one convex hull, SSVT utilizes the
underlying relationship of the target candidates. Moreover, SSVT is very efficient in that
it only needs to solve one quadratic optimization problem rather than sparse optimization
problems. Both qualitative and quantitative analyses on several challenging image sequences show that the proposed SSVT algorithm outperforms the state-of-the-art trackers.
Keywords: set to set distance, visual tracking, particle lter, convex hull, support vector machine
155
Setting An Effective Pricing Policy for Double Auction Marketplaces
Bing Shi1, Yalong Huang1, Shengwu Xiong1, and Enrico H. Gerding2
1School of Computer Science and Technology, Wuhan University of Technology,
Wuhan, China, [email protected]
2School of Electronics and Computer Science,
University of Southampton, Southampton, UK
Abstract. In this paper, we analyse how double auction marketplaces set an effective pricing policy to determine the transaction prices for matched buyers and sellers. We analyse
this problem by considering continuous privately known trader types. Furthermore, we
consider two typical pricing policies: equilibrium k pricing policy and discriminatory k
pricing policy. We firstly investigate how to determine the transaction prices to reach the
maximal allocative efficiency in an isolated marketplace when the traders adopt BayesNash equilibrium bidding strategies. We find that when the marketplace adopts discriminatory k pricing policy, the maximal allocative efficiency is reached by setting k=0.41
or 0.59.We find that equilibrium k pricing policy provides higher allocative efficiency
than discriminatory k pricing policy.We further discuss how different pricing policies
can affect traders’ Bayes-Nash equilibrium bidding strategies. Furthermore, we extend
the analysis to the setting with two marketplaces competing against each other to attract
traders. We find that the marketplace using equilibrium k pricing policy is more likely
to beat the marketplace using discriminatory k pricing policy, where all traders converge to the marketplace using equilibrium k pricing policy in Bayes-Nash equilibrium.
Our analysis can provide meaningful insights for designing an effective pricing policy.
Keywords: Pricing Policy, Double Auction, Market Selection and Bidding Strategy,
Fictitious Play, Bayes-Nash Equilibrium
156
Single Image Super-Resolution based on Nonlocal Sparse and
Low-Rank Regularization
Chunhong Liu, Faming Fang, Yingying Xu, and Chaomin Shen*
Shanghai Key Laboratory of Multidimensional Information Processing
East China Normal University
Shanghai 200241, China
[email protected]
Abstract. Image super resolution (SR) is an active research topic to obtain an high
resolution (HR) image from the low resolution (LR) observation. Many results of existing methods may be corrupted by some artifacts. In this paper, we propose an SR
reconstruction method for single image based on nonlocal sparse and low-rank regularization. We form a matrix for each patch with its vectorized similar patches to
utilize the redundancy of similar patches in natural images. This matrix can be decomposed as the low rank component and sparse part, where the low rank component depictures the similarity and the sparse part depictures the ne dierences and
outliers. The SR result is achieved by the iterative method and corroborated by experimental results, showing that our method outperforms other prevalent methods.
Keywords: Super resolution, low-rank, sparsity, nonlocal self-similarity.
Stigmergy-based Influence Maximization in Social Networks
Weihua Li1, Quan Bai1, Chang Jiang1, and Minjie Zhang2
1Auckland University of Technology, New Zealand,
[email protected]; [email protected]; [email protected]
2University of Wollongong, Australia
[email protected]
Abstract. Influence maximization is an important research topic which has been extensively studied in various elds. In this paper, a stigmergy based approach has been proposed to tackle the influence maximization problem. We modelled the influence propagation process as ant’s crawling behaviours, and their communications rely on a kind of
biological chemicals, i.e., pheromone. The amount of the pheromone allocation is concerning the factors of influence propagation in the social network. The model is capable
of analysing influential relationships in a social network in decentralized manners and
identifying the influential users more efficiently than traditional seed selection algorithms.
Keywords: Influence maximization, ant algorithm, stigmergy
157
SWARM: An Approach for Mining Semantic Association Rules from
Semantic Web Data
Molood Barati1, Quan Bai1, and Qing Liu2
1Auckland University of Technology, New Zealand,
2The Commonwealth Scientic and Industrial Research Organization (CSIRO), Australia
[email protected]; [email protected]; [email protected]
Abstract. The ever growing amount of Semantic Web data has made it increasingly dicult to analyse the information required by the users. Association rule mining is
one of the most useful techniques for discovering frequent patterns among RDF triples. In this context, some statistical methods strongly rely on the user intervention
that is time-consuming and error-prone due to a large amount of data. In these studies,
the rule quality factors (e.g. Support and Condence measures) consider only knowledge in the instance-level data. However, Semantic Web data contains knowledge in
both instance-level and schema-level. In this paper, we introduce an approach called
SWARM (Semantic Web Association Rule Mining) to automatically mine Semantic
Association Rules from RDF data. We discuss how to utilize knowledge encode in the
schema-level to enrich the semantics of rules. We also show that our approach is able
to reveal common behavioral patterns associated with knowledge in the instance-level
and schema-level. The proposed rule quality factors (Support and Condence) consider
knowledge not only in the instance-level but also schema-level. Experiments performed
on the DBpedia Dataset (3.8) demonstrate the usefulness of the proposed approach.
Keywords: Semantic Web data, Association Rule Mining, Ontology, Knowledge
Discovery
158
Thai Printed Character Recognition Using Long Short-Term Memory
and Vertical Component Shifting
Taweesak Emsawas, Boonserm Kijsirikul
Department of Computer Engineering, Chulalongkorn University,
Phayathai Rd., Phathumwan, Bangkok, 10330, Thailand
Abstract. The segmentation-based approach for Optical Character Recognition (OCR)
works by first segmenting a text line image into individual character images and then recognizing the characters. The approach relies heavily on the performance of the segmentation process and thus suffers from the problem of touching and broken characters. On
the other hand, the unsegmented approach for OCR processes the text line image without segmenting the image into individual characters, and the approach is more suitable
for languages such as Thai that contains a lot of touching characters in nature. This paper
proposes an application of Long Short-Term Memory (LSTM), which is an unsegmented method, to Thai OCR. The paper also introduces a method called vertical com-ponent
shifting to solve the problem of a large number of vertically occurring character combinations that occur in four-level writing system of Thai, and pose difficulty for standard
LSTM networks. The experimental results demonstrate the better accuracy of our proposed
method over standard LSTM networks and other commercial software for Thai OCR.
Keywords: Thai Printed Character Recognition, Recurrent Neural Network, Long
Short-Term Memory, Vertical Component Shifting
159
Threshold-based Direct Computation of Skyline Objects for Database
with Uncertain Preferences
Venkateswara Rao Kagita1, Arun K Pujari12, Vineet Padmanabhan1, Vikas
Kumar1 and Sandeep Kumar Sahu1
1School of Computer and Information Sciences, University of Hyderabad, India.
2Central University of Rajasthan, India.
[email protected], {akpcs,vineetcs}@uohyd.ernet.in,
{vikas,uusandeepsahu}@uohyd.ac.in
Abstract. Skyline queries aim at nding a set of skyline objects from the given database. For categorical data, the notion of preferences is used to determine skyline objects. There are many real world applications where the preference can be uncertain.
In such contexts, it is relevant to determine the probability that an object is a skyline
object in a database with uncertain pairwise preferences. Skyline query is to determine
a set of objects having skyline probability greater than a threshold. In this paper, we
address this problem. To the best of our knowledge, there has not been any technique
which handles this problem directly. There have been proposals to compute skyline
probability of individual objects but applying these for skyline query is computationally expensive. In this paper, we propose a holistic algorithm that determines the set
of skyline objects for a given threshold and a database of uncertain preferences. We
establish the relationship between skyline probability and the probability of the union
of events. We guide our search to prune objects which are unlikely to be skyline objects.
We report extensive experimental analysis to justify the efficiency of our algorithm.
160
Topic Detection in Group Chat Based on Implicit Reply
Xinyu Zhang, Ning Zheng, Jian Xu, and Ming Xu
School of Computer Science and Technology, Hangzhou Dianzi University
Hangzhou, China
{132050125,nzheng,jian.xu,mxu}@hdu.edu.cn
Abstract. Topic detection in group chat has become a promising research due to
the widely usage of Instant Messaging (IM) systems. Previous works mainly focus
on improving the text similarity between two related messages by utilizing dierent
weighting factors. However, the text similarity of related texts is likely to be zero (or
near zero) due to the characteristics of short text messages in group chat. To solve
this problem, an innovative topic detection method based on implicit reply which
indicates chat messages interact with each other is proposed in this paper. The comparative experiments results on the datasets gathered from QQ groups demonstrate
the superiority of the proposed method as compared to the baseline approaches.
Keywords: Topic detection, Group chat, Multi-topic Window
Towards Exposing Cyberstalkers in Online Social Networks
Jiamou Liu1, Yingying Tao2, and Quan Bai2
1The University of Auckland, New Zealand
[email protected]
2Auckland University of Technology
{kdx8538,quan.bai}@aut.ac.nz
Abstract. This paper presents work-in-progress towards a computational approach for
capturing a type of deviant behaviors, that are characterised by persistent monitoring
and information gathering through online social medias. Such behaviors are co monly
associated with stalking on the cyberspace. We present a network-based framework for
describing online user interactions. Based on this framework we provide a description
of excessive and unreciprocated attention an agent pays to another agent. We conclude
with a discussion on limitation of the current work and a guideline for future extensions.
Keywords: Cyberstalking online social network agent behaviors
161
Using Canonical Correlation Analysis For Parallelized
Attribute Reduction
Ping Li, Mengting Xu, Jianyang Wu, Lin Shang
State Key Laboratory for Novel Software Technology,
[email protected]
[email protected]
[email protected]
[email protected]
Abstract. Attribute reduction in rough sets theory has been widely used in classication.
Classical attribute reduction algorithm only considers correlation between condition
attributes and decision attributes, which ignores the relationship among condition attributes themselves. Moreover, when faced with large-scale data, running time of classical
attribute reduction algorithm has been increasing. Aiming to solve these two problems,
a parallelized reduction algorithm called P-CCARough Reduction is proposed in this
paper. The algorithm employs canonical correlation analysis named CCAFusion and
parallelized attribute reduction algorithm named P RoughReduction. CCAFusion divides the original set of attributes into two subsets randomly. Then the correlations of
these two subsets of features are analyzed. After that, the attributes are fused into one
collection according to the derived correlations. P-RoughReduction algorithm is based
on a distributed framework MapReduce which parallelizes the classical attribute reduction algorithm according to the attribute importance in rough sets theory. It is shown
that P-CCARoughReduction algorithm through experiments on 50000 samples not only
performs well on time, the classication accuracy has also been signicantly improved.
Keywords: Canonical correlation analysis, Rough sets theory, Attribute reduction,
MapReduce.
162
PRICAI Workshop
(AI4T, AIED, I3A, IWEC, RSAI, PeHealth)
Abstracts
163
A Framework to Generate Carrier Path Using Semantic Similarity of
Competencies in Job Position
Wasan Na Chai1, Taneth Ruangrajitpakorn12,
Marut Buranarach1, and Thepchai Supnithi1
1Language and Semantic Techonolgy Laboratory,
National Electronics and Computer Technology Center, Pathumthanee, Thailand
{wasan.na_chai,taneth.rua,marut.bur,thepchai.sup}@nectec.or.th
2Department of Computer Science, Faculty of Science and Technology,
Thammasat University, Pathumthanee, Thailand
Abstract. A career path is necessary for students and workers to keep themselves in
track for their career goal. However, a career path following a job standard in general is very rare. This paper presents a method to find a semantic similarity within
competencies of Job positions for realising a path to relate career. By a development of Thai WordNet containing terms used in competency description, a distance
of classes of WordNet structure is used to determine a semantic similarity of comp
tencies. Paths to relate job positions are assumed for the job positions sharing similar competencies, and the more they share, the more transferrable job is viable.
From the usage scenario, the proposed framework proved that semantic of words
is more useful than using character based similarity in competency comparison.
Keywords: Career Path * Semantic Similarity * WordNet * Competency
164
A multi-objective adaptive Invasive Weed Optimization intelligence
approach for solving DNA sequence design
Qiang Zhang*, Gaijing Yang, Changjun Zhou, Bin Wang
Key Laboratory of Advanced Design and Intelligent Computing (Dalian university),
Ministry of Education, Dalian, 116622, China
[email protected]
Abstract. DNA sequence design is a key factor to ensure the success of DNA computing. In order to make the DNA computing more reliable, many studies have focused
on the DNA sequence design. DNA sequence design relates to several and conflicting
design criteria, in this paper, DNA sequence design is formulated as a multi-objective
optimization problem, and solved by using a multi-objective adaptive Invasive Weed
Optimization intelligence approach (MA_IWO for short). Concretely speaking, our approach (MA_IWO) generates reliable DNA sequences with the consideration of six different conflicting design criteria by introducing the fast non-dominated sorting. Moreover, adaptive mechanism is introduced into the reproduction phase of the Invasive
Weed Optimization algorithm (IWO for short), so that the standard deviation of each
generation can be changed adaptively according to the fitness value. In addition, our
results are validated by comparison with other related works published in the literature.
What can be concluded is that the novel approach presented in the paper obtains very
satisfactory results which significantly surpass the other previously published results.
Keywords: DNA sequence design; DNA computing; IWO algorithm; fast nondominated sorting.
165
A Regression-based SVD Parallelization using Overlapping
Folds for Textual Data
Uraiwan Buatoom, Thanaruk Theeramunkong, and Waree Kongprewechnon
School of Information, Communication and Technology,
Sirindhorn International Institute ofTechnology, Thammasat University, Thailand
Abstract. One of the most difficult issues in text mining is high dimensionality caused
by a large number of features (keywords). While various multivariate analyses, such
as PCA and SVD (in information retrieval, called LSI), are developed to solve this
curse of high dimensionality, they are computationally costly. This paper investigates a regression-based reconstruction method that enables parallelization of PCA/
SVD by decomposing a document-term matrix into a set of sub-matrices with consideration of overlapped terms, and then to re-assemble using regression technique.
To evaluate our method, we utilize two text datasets in the UCI Machine Learning
Repository, called “Bag of Words” and “Reuter 50 50”. To measure the closeness
between two documents, cosine similarity is applied while the accuracy is measured in the form of rank order mismatch. Finally, the result shows that, the matrices decomposition and reassembly can preserve the quality of relation/representation.
Keywords: Decomposition, Re-assembly, Sub-matrix, SVD, Regression, LSI
166
Affective Laughter Expressions from Body Movements
Ma. Beatrice L. Luz, McAnjelo D. Nocum, Timothy Jasper T. Purganan,
Wing San T. Wong, Jocelynn W. Cu
Center for Human Computing Innovations, De La Salle University, Philippines
{ma_beatrice_luz, mcanjelo_nocum, timothy_purganan,
wing_wong, jocelynn.cu}@dlsu.edu.ph
Abstract. The main goal of this study is to classify affective laughter expressions
from body movements. Using a non-intrusive Kinect sensor, body movement data
from laughing participants were collected, annotated and segmented. A set of features that include the head, torso, shoulder movements, as well as the positions
of the right and left hands, were used by a decision tree classifier to determine the
type of emotions expressed in the laughter. The decision tree classifier performed
with an accuracy of 71.02% using a minimum set of body movement features.
Keywords: affective laughter, laughter expression, analysis of body movement, gestures
167
Application of Annotation on Smoothing for Subject-independent
Emotion Recognition based on Electroencephalogram
Nattapong Thammasan1, Ken-ichi Fukui2, and Masayuki Numao2
1Graduate School of Information Science and Technology, Osaka University
Suita-shi, Osaka 565-0871, Japan
[email protected]
2Institute of Scientic and Industrial Research (ISIR), Osaka University
Ibaraki-shi, Osaka 567-0047, Japan
{fukui,numao}@ai.sanken.osaka-u.ac.jp
Abstract. In the construction of computational models to recognize emotional state,
emotion reporting continuously in time is essential based on the assumption that emotional responses of a human to certain stimuli could vary over time. However, currently
existing methods to annotate emotion in temporal continuous fashion are confronting
various types of challenges. Therefore, the manipulation of the annotated emotion prior
to labeling training samples is necessary. In this work, we present an early attempt to
manipulate the emotion annotated in arousal-valence space by applying three dierent
signal ltering techniques to smooth annotation data; moving average lter, Savitzky-Golay lter, and median lter. We conducted experiments of emotion recognition in music
listening tasks employing brainwave signals recorded from an electroencephalogram
(EEG). Smoothed annotation data were used to label the features extracted from EEG
signals to train emotion recognizers using classication and regression techniques. Our
empirical results indicated the potential of the moving average lter that could increase
the performance of emotion recognition evaluated in subject-independent fashion.
Keywords: Emotion recognition, Electroencephalogram, Music-emotion, Annotation,
Smoothing
168
Arrival Time Prediction and Train Tracking Analysis
Arrival Time Prediction and Train Tracking Analysis
Somkiat Kosolsombat, and Wasit Limprasert
Department of Computer Science, Faculty of Science and Technology,
Thammasat University, Pathumthani, Thailand
{somkiat.k,wasit_l}@sci.tu.ac.th
Abstract. Rail transportation is a convenient and safe in many countries. However,
Rail transportation in some countries has significant long delays. Arrival time prdiction and rescheduling the time table are partial solutions to tackle the delay problem.
In this paper, the relationship between measurable properties and the delay time are
studied in order to develop an arrival time prediction. The result of this experiment
has three parts. The relationship between properties and arrival late are then visualized and discussed. Some properties from the acquired database show that week,
day and station, are important features and impact on the delay. Various regression methods are compared in our experiment and the result shows that best RMSE
is ±3.863 minutes by applying Random Forest Regression on train tracking dataset.
Keywords: arrival time prediction, train tracking analysis, arrival regression
169
Automatic Question Generation on SQL Language Using
Template-Based Method
Jittima Janphat1 and Orawan Chaowalit2
1Department of Computing, Faculty of Science, Silpakorn University,
Nakhon Pathom, Thailand
[email protected]
2Department of Computing, Faculty of Science, Silpakorn University,
Nakhon Pathom, Thailand
[email protected]
Abstract. The objective of this research is to generate database questions auto-matically. Creating questions for learners to write SQL statement either to practice or to
assess their knowledge level is time consuming for instructors. The system can reduce
this workload by automatically generates questions for instructors. Template formats
were created from sample exercises from database textbooks, and then the templates
were filled with data from metadata about databases and data in database, which is
to be put in the database management system by the instructor. After that, the system
generates SQL question corre-sponding to the database to facilitate the instructor. Four
experts from Silpakorn university who are familiar with database subject evaluated the
reliability and the level of learning of questions that were generated by the system.
Keywords: Question generation · Metadata · Question template · SQL
170
Building a Semantic Ontology for Virtual Peers in
Narrative-Based Environments
1Ethel Chua Joy Ong, Danielle Grace Consignado,
Sabrina Jane Ong and Zhayne Chong Soriano
Center for Language Technologies, De La Salle University, Manila, Philippines
[email protected]
Abstract. Narrative-based environments utilize various forms of knowledge to provide an interactive space for the learner and the virtual agent to collaborate in accomplishing the learning goals. In this paper, we present the design of a semantic ontology that provides the necessary domain-based conceptual knowledge to
allow a virtual peer to engage in storytelling as a form of exchange with the learner. We then show how the ontology was utilized to support the virtual peer in performing its tasks, which include generating interactive stories that teach about apropriate social behavior, and engaging in a textbased dialogue with the learner.
Keywords: Semantic Ontology, Virtual Peer, Story Generation, Dialogue Generation,
Commonsense Knowledge
171
Classification of Diabetic Retinopathy Stages using Image Segmentation
and an Artificial Neural Network
Narit Hnoohom1 and Ratikanlaya Tanthuwapathom2
Image, Information and Intelligence Laboratory,
Department of Computer Engineering,
Faculty of Engineering, Mahidol University, NakornPathom, Thailand
Abstract. Diabetic retinopathy, which can lead to blindness, has been found in 22 percent of diabetic patients in the latest survey. Therefore, diabetic patients should have an
eye examination at least once a year. However, it has been found that currently there is
a problematic lack of specialists in ophthalmology. Detection and treatment of diabetic
retinopathy are thus delayed. The idea to create a classification system of diabetic retinopathy stages to facilitate the making of preliminary decisions by ophthalmologists
is introduced. This paper presents the classification of diabetic retinopathy stages using
image segmentation and an artificial neural network. This proposed method applies local
thresholding to separate the foreground region from the background region so that the
optic disc and exudate regions are able to be identified more clearly. The experiment was
carried out with 100 fundus images from the Institute of Medical Research and Technology Assessment database. The prediction model had an accuracy rate of up to 96 percent.
Keywords: Diabetic retinopathy, Exudates, Fundus image, Artificial neural network.
172
Comparison of Edge Detection Algorithms for Coastline Detection in
Satellite Imageries
Chutiwan Boonarchatong1, Sucha Smanchat2,
Mahasak Ketcham2, and Nawaporn Wisitpongphan2
1Department of Information Technology, Suan Dusit University,
Thailand [email protected]
2Department of Information Technology,
King Mongkut’s University of Technology
North Bangkok, Thailand
[email protected], {mahasak.k, nawaporn.w}@ it.kmutnb.ac.th
Abstract. Finding the edge of a coastline is one of the crucial tasks in monitoring the
coastline erosion which is an indicator of ecological change. The goal of this work is
to find a suitable edge detection operator. In this work, the dataset was derived from
the raw satellite imageries, namely THEOS. The result shows that Canny and Laplacian of Gaussian are the best detector in both less noise and Medium noise when compared with Robert, Sobel, Prewitt, and Laplacian of Gaussian detection algorithms.
Keywords: edge detection • coastline • satellite imagery
173
Computational Model for Affect Detection in Learning
Najlaa Sadiq Mokhtar and Syaheerah Lebai Lut
School of Computer Sciences,
Universiti Sains Malaysia,
11800, Pulau Pinang, Malaysia
Abstract. Generally, computer-based tutoring system is less engaging compared to
human tutoring due to its “insensitiveness” to learners’ affect. A truly intelligent tutoring system (ITS) will take into consideration the learner’s state, and adapt to that
state before providing an appropriate learning content or feedback to improve learning.
However, an ITS needs to rstly understand the learner’s state and his/her environment.
This step is crucial to give appropriate feedback. This study focused on the four (4)
frequent emotions attached to learning, which are frustration, boredom, uncertainty and
neutral. Using several existing task-based features from previous studies centered in
the west, combined with new features from our study, we constructed a four-class localized computational model of affect detection in learning through machine learning
approach. The features collected were evaluated with several standard classiers. Results revealed that the J48 classier learned best when evaluated using the task-based
features using a host e-tutorial that was especially developed for evaluation purposes.
Keywords: Intelligent tutoring system; computational emotion model; affect detection; learning; culture-sensitive
174
Contents Organization Support for Logical Presentation Flow
Tomoko KOJIRI1 and Yuta WATANABE2
1Faculty of Engineering Science, Kansai University
2Graduate School of Science and Engineering, Kansai University
3-3-35, Yamate-cho, Suita, Osaka, 5648680 Japan
[email protected]
Abstract. In appropriate presentations, each topic must be explained logically. To prepare appropriate presentation slides, all of the topics related to the research must be fully elucidated. The objective of this research is to help presenters
derive enough topics that cogently explain their research theme. This paper proposes the logical model for the topics in the research presentation of the computer science field. Then, content organization support system that helps presenters organize topics based on the logical model is developed. Based on the experiment, our
system was effective for creating a new contents and for organizing contents.
Keywords: content map, causal relation, presentation support, logical presentation
175
Desktop Tower Defense is NP-Hard
Vasin Suttichaya
Faculty of Engineering, Mahidol University,
999 Phuttamonthon 4 Road, Salaya 73170, Thailand,
[email protected]
Abstract. This paper proves the hardness of the Desktop Tower Defense game. Specically, the problem of determining where to locate k turrets in the grid of size m n in order to
maximize the minimum distance from the starting point to the terminating point is shown
to be NP-hard. The proof applies to the generalized version of the Desktop Tower Defense.
Keywords: Graph theory, Hamiltonian path, Complexity
176
Development of Salary Prediction System to Improve Student Motivation
using Data Mining Technique
Pornthep Khongchai1, Pokpong Songmuang2
12Department of Computer Science, Faculty of Science and Technology,
Abstract. This paper aimed to determine an efficient data mining technique for salary
prediction to motivate the eagerness to study. Five data mining techniques were compared: Decision trees, Naive Bayes, K-Nearest neighbor, Support vector machines, and
Neural networks. To evaluate the relative efficiencies of the techniques, 13,541 records
of graduated student data were used in 10-fold cross validation. Results showed that
K-Nearest neighbor provided the best efficiency. K-Nearest neigh-bor was also applied
as a model for salary prediction. A questionnaire survey was used to evaluate the effectiveness of the system with 50 student samples. Results indicated that the system was
effective in boosting students’ motivation for studying and also gave them a positive
future viewpoint. The student sample registered positive satisfaction in using the system, since it was easy to use and the predictive results were simple and comprehensible.
Keywords: Educational Data Mining, Classification technique, Decision trees, Naïve
Bayes, K-Nearest neighbor, Support Vector Machines, Neural Networks
177
Examination Timetabling using Prey Predator Algorithm
Surafel Luleseged Tilahun and Jean Medard T Ngnotchouye
School of Mathematics, Statistics and Computer Science,
University of KwaZulu-Natal, 3209, Pietermaritzburg, South Africa,
[email protected]
Abstract. Education timetabling problem is a problem of arranging exams so that there
will not be a clash with the objective of using a minimum number of time slots. It is a
combinatorial optimization problem which has been studied by different authors. For
combinatorial problems in general, and educational timetabling problem in particular,
different metaheuristic algorithms have been proposed and used. Some of these algorithms are originally proposed for continuous problems and later extended for combinatorial optimization problems and are found to have a promising performance. Based on
the well-known ‘no free lunch theorem’, there is no superior algorithm for all problem
domains. This implies that if an algorithm performs better than another algorithm in
some problems or problem domain then there exist another set of problems or problem domains in which it will be out performed. Hence, having different approaches
to deal with these problems is advantageous, while how to choose an algorithm for a
problem at hand remains to be a research to be studied further. Hence, in this paper,
prey predator algorithm will be modified to suit combinatorial optimization problems
in general, and educational exam timetabling in particular, with adapted local search
mechanism which guides the search over the constrained set of solutions. In order to
test the approach five problem instances are chosen from Carter uncapacitated exam
timetabling benchmark problem. These problem are well known benchmark problems
for high dimensional combinatorial optimization problems. Simulation results show
that the proposed approach is promising and comparable with previous approaches.
Keywords: Educational timetabling, prey predator algorithm (PPA), metaheuristic
178
Enhancement of Palm-Leaf Manuscript for Segmentation
Siriya Phattarachairawee, Montean Rattanasiriwongwut, Mahasak Ketcham
Department of Information Technology Faculty of Information Technology King
Mongkut’s University of Technology North Bangkok, Thailand
[email protected]
[email protected]
[email protected]
Abstract. This paper presents Enhancement of Palm-Leaf Manuscript as noise occurred in images It’s suitable for Segmentation. The proposed method consists of image
adjustment, contrast stretching and histogram equalization technique. All these techniques improve the quality of images for human viewing. The method we proposed,
it made more effective images of palm-leaf manuscripts. The color image of palm-leaf
manuscripts became pale and the visible images are not beautiful and hardly colorless
because of image enhancement in image processing. In the other hand, the proposed
method enhanced clarity of black alphabets. Thus, the color images were transformed
into grayscale images affects the clearest alphabets on palm-leaf manuscripts. It also reduces noise. Next, the output images are suitable for carrying out a recognition system.
Keyword: Palm-Leaf Manuscript, Adjustment, Stretching, Histogram
179
Estimating PSD Characteristics of ECG in Comparison between Normal
and Supraventricular Subjects
Thaweesak Yingthawornsuk, Siriphan Phetnuam,
Saowaros Singkhal, Waraporn Pattarason
Media Technology, King Mongkut’s University of Technology Thonburi, Thailand
[email protected]
Abstract. The aims of project are to develop an arithmetic program that can detect
irregularity in electrocardiogram (ECG) and classify between two groups of normal and
supraventricular ECG waveforms by using Auto regressive (AR) estimators with various model orders starting from 3rd to 9th. All AR estimators are associated with the PSD
of ECG waveforms collected from a group of 30 subjects at 200Hz sampling frequency.
The best classification scores found on the 5th-order AR model are 95.99% and 72.17%
obtained from training and testing the C4_5 classifier with the fifth-order coefficients.
By classifying the 7th-order AR coefficients with Linear Least Squared (LS) classifier
the accurate scores of 86.43% and 80.85% were obtained from training and testing cases respectively. These performance accuracies show that the proposed method is highly
effective in parameterizing and classifying PSD feature as quantitative measure that
can characterize the ECG signals of normal and supraventricular cardiac conditions.
Keywords: ECG, PSD, AR, Supraventricular
180
Evolving Public Opinion Mining Methods on Decision Support System in
Thai E-Government
Jeerana Noymanee1, Wimol San-Um2 and Thanaruk Theeramunkong3
1Electronic Government Agency (Public Organization) of Thailand ,
Bangkok, Thailand
2Intelligent Electronic System Research Laboratory,
Thai-Nichi Institute of Technology, Bangkok, Thailand
3School of Information, Computer, and Communication Technology (ICT) Sirindhorn
International Institute of Technology, Thammasat University P.O.Box 22,
Pathum Thani 12121, Thailand.
Jeerana @ega.or.th, [email protected], [email protected]
Abstract. Public opinion mining is a combination of Natural Language Processing
(NLP) and Sentiment Analysis. To make appreciate decisions in policy, it is necessary
to utilize sentiment classification efficiently. While reviews usually contain sentiment
which is expressed in a different way in different domains, it is costly to annotate data
for each new domain. E-Government refers to the use of information and communications technologies (ICT) to improve quality of services and information offered to
citizens, and government in order to obtain more accountable and transparency towards
governance in public sector. Recently, have been widely discussed governmental decisions within digital societies. This paper provide and exploration of opinion mining
and text mining techniques towards apprehending the public’s opinion communicated online and concerning governmental decisions. Regarding the objective of study is
focuses on the understanding of the citizen opinions about e-Government issues and
on the exploitation of these opinions in subsequent governmental actions. This paper
also examine several features in the user generated content discussing governmental
decisions in an attempt to automatically extract the citizen opinions from online posts
on public sector regulations and thereafter it can be able to organize the extracted opinions not only into polarized clusters but also collect the potential word that able to
declare the citizen demand. The objective is to identify the public’s stance against governmental decisions automatically and It can be deduced that how the citizen’s attitudes
may effect to government actions. To demonstrate the usability and added value of the
proposed the architecture of e-Government will be presented and discussed in paper.
Keywords: Opinion mining, opinion classification, E-Government, Decision support
system, Online Social Network
181
Exploring the Distributional Semantic Relation for ADR and Therapeutic
Indication Identification in EMR
Siriwon Taewijit12 and Thanaruk Theeramunkong1
1Sirindhorn International Institute of Technology, Thammasat University,
Pathum Thani, Thailand
2Japan Advanced Institute of Science and Technology, Ishikawa, Japan
Abstract. Extraction of relations and their semantic relations from a clinical text is
significant to comprehend the actionable harmful and beneficial events between two
clinical entities. Particularly to implement drug safety surveillance, two simplest but
most important semantic relations are adverse drug reaction and therapeutic indication. In this paper, a method to identify such semantic relations is proposed. A large
scale of nearly 1.6 million sentences over 50,998 discharge summary from Electronic Medical Records were preliminary explored. Our approach provided the three
main contributions; (i) Electronic Medical Records characteristic exploration; (ii)
OpenIE examination for clinical text mining; (iii) automatic semantic relation identification. In this paper, the two complementary information from public knowledge
base were introduced as a comparative advantage over expert annotation. Then the
set of relation patterns were qualified with 0.05 significant level. The experimental results show that our method can identify the common adverse drug reaction and
therapeutic indication with the high lift value. Additionally, a novel adverse drug reaction and alternative drug for a specific symptom therapy are reported to support
the comprehensive further drug safety surveillance. The paper clearly illustrates that
our method is not only effortless from expert annotation, automatic pattern–specific semantic relation extraction, but also effective for semantic relation identification.
Keywords: adverse drug reaction, electronic medical records, semantic relation extraction, therapeutic indication, text mining
182
Extracting and Characterizing Functional Communities
in Spatial Networks
Takayasu Fushimi1, Kazumi Saito2, Tetsuo Ikeda2, and Kazuhiro Kazama3
1University of Tsukuba, Japan, [email protected]
2University of Shizuoka, Japan, fk-saito,[email protected]
3Wakayama University, Japan, [email protected]
Abstract. We address the problem of extracting and characterizing functional communities consisting of functional similar regions in spatial networks such as urban
streets. Such characteristics of regions will play important roles for developing and
planning city promotion, travel tours and so on, as well as understanding and improving the usage of urban streets. In order to analyze such functionally similar regions,
based on a previous algorithm for extracting functional communities for each network,
we propose a new method consisting of a technique for simultaneously comparing
these functional communities of several networks, and an effective way of visualizing these communities calculated from OpenStreetMap data, by especially focusing
on a fact that the maximum degree of nodes in spatial networks is restricted to relatively small numbers. In our experiments using urban streets of six cities, we show
that our method can produce a series of useful visualization results accompanied with
interpretable functional communities. Moreover, we empirically conrm that our results
are substantially different from those obtained by representative centrality measures.
183
Fatigue Classification of Military Mission by EEG signals
via Artificial Neural Network (ANN)
Worawut Yimyam1, and Mahasak Ketcham2
1Department of Computer Business, Phetchaburi Rajabhat University, Thailand
[email protected]
2Department of Information Technology Management,
king Mongkut’s University of Technology North Bangkok, Thailand
[email protected]
Abstract. This paper proposes the development of an algorithm used for monitoring fatigue of the soldiers while they perform their duty. Electroencephalography (EEG) signals are analyzed by an Artificial Neural Networks (ANN) technique and compared with other techniques. The experimental results show that
the ANN provides more accurate results than Bayesnet, Support Vector Machines
(SMO), and Naïve Bayes techniques. The result of the ANN technique provides
the accuracy, recall, and precision values at 83.77, 0.838, and 0.838, respectively.
Keywords: Fatigue, Electroencephalography, Artificial Neural Networks
184
Inferring Tourist Behavior and Purposes of a Twitter User
Yuya Nozawa1,*, Masaki Endo12, Yo Ehara3, Masaharu Hirota4,
Syohei Yokoyama1, and Hiroshi Ishikawa1
1Graduate School of System Design, Tokyo Metropolitan University, Japan
{nozawa-yuya,endo-masaki}@ed.tmu.ac.jp, [email protected]
2Division of Core Manufacturing, Polytechnic University, Japan
3National Institute of Advanced Industrial Science and Technology, Japan
[email protected]
4Department of Information Engineering,
National Institute of Technology, Oita College, Japan
[email protected]
5Faculty of Informatics, Shizuoka University, Hamamatsu, Japan
[email protected]
Abstract. The importance of tourism information such as tourism purposes and tourist behavior continues to increase. However, obtaining precise tourist information
such as the tourist destination and tourism period is difficult, as is applying that information to actual tourism marketing. We propose a method to classify Twitter
user into tourist behavior and tourism purposes, extracting related information from
Twitter posts. Our experiments demonstrated a 0.65 F-score for multi-class classification, showing accuracy for inferring tourist behavior and tourism purposes.
Keywords: Attribute estimation, Travel information
185
K-Mean Algorithm for Finding Students’ Proficiency with
a Framework’s item Examination
Nongnuch Ketui1, Kanitha Homjun2, and Prasert Luegkhong3
1Computer Science Program, Faculty of Sciences and Agricultural Technology
Rajamangala University of Technology Lanna, Nan, Thailand
2Information Technology Program, Faculty of Sciences and Agricultural Technology
Rajamangala University of Technology Lanna, Nan, Thailand
3College of Integrated Science and Technology,
Rajamangala University of Technology Lanna, Chiang Mai, Thailand
{nongnuchketui,kanithaasc,prasert}@rmutl.ac.th
Abstract. The effective system of item examination is used to assess the achievement
students while the prociency of students can be classify theirs real aptitude. In this
paper, we introduce a framework of item examination which having two phases; (1)
building and analyzing item examination and (2) finding the student’s prociency. To
analyze the online item, a number of experiments are conducted using 1,000 items of
Information Technology (IT) which having four choices in each item and evaluated by
two groups of students. The questions that having the right answer are segmented into
a small unit (IT word), assigned with the frequency weighting, and clustered into six
groups of IT aptitudes. Based on measures of KR-20 called the reliability value and
evaluated the performance of item examination three factors; diculty index, discrimination values, and distracter efficiency. While the student’s prociency is explored by
assigning IT keywords with the standard weighting (TF/IDF) in order to clustering with
K-Mean clustering algorithm. The experimental results show that the reliability is good
level (0.98) and the performance of students are quite good in information technology.
Keywords: item examination, student prociency, K-Mean algorithm
186
Learning Latent Word Representations for
Enhanced Short Text Classification
Luepol Pipanmaekaporn1 and Suwatchai Kamolsantiroj2
Department of Computer and Information Science,
King Mongkut’s University of Technology North Bangkok,
Bangkok, Thailand 10800
E-mail:{luepolp1,suwatchaik2}@sci.kmutnb.ac.th
Abstract. Web short texts have been increasingly available in the past few years but
conventional approaches to text classification are not suitable for short texts due to
the data sparseness problem. In this work, we propose a novel rep-resentation learning method to tackle this challenge. Our key idea is to learn re-liable low-dimensional dense representations for short text data based on latent word representations that
capture semantics of words over corpus. To efficiently build the word representations,
we first compute term similarity based on Word2vec, a deep learning tool that learns
semantic vectors of words. We then learn a latent space of individual words from term
similarity information using sparse autoencoder. By using the latent word space, we
learn feature vectors for short documents based on error minimization. We conduct
experiments on the two classification tasks: sentiment text classification and news title classifi-cation to evaluate the proposed method. Experimental results on two real-world datasets demonstrate that our proposed method produces more stable features
that enhance short-text classification than state-of-the-art latent feature repre-sentations.
Keywords: short text classification, document representation, representation learning,
latent features and sparse autoencoder.
187
Medicine Recognition using Intrinsic Geometric Property from Pill Image
Md. Zakir Hossan, Tanjina Piash Proma, M. Ashraful Amin
Computer Vision and Cybernetics Group,
Department of Computer Science and Engineering,
Independent University, Bangladesh, Bashundhora R/A, Dhaka, Bangladesh
Abstract. It is often the case that prescription pills do not come with blister or alu
alu packaging where the identity of the pill is available, rather it comes in air-tight
plastic bottles or labeled Ziploc bags. Problem with such bottle or pack is that, if
the label is removed then it becomes difficult to tell what the pill is. Moreover,
there is the issue of visually impaired people having difficulty identifying pills outside the pack. There are such many scenarios where it is good to have an automated
pill recognition system. Due to the large variety of size, shape, color, texture it is a
difficult task for human to tell about the identity of any individual medical pill. To
localize a pill from a given dataset using computer vision techniques requires multiple steps. This paper will describe how to split a dataset according to the shape of
pill. To find the shape information we used intrinsic geometric properties such as:
eccentricity, extent and narrowness of pill which can be extracted from image using
carefully selected image processing techniques. Reference values of discriminative
parameters are determined using ‘RxIMAGE’, National Library of Medicine, USA database. The overall shape discrimination accuracy of the proposed system is 93.75%.
Keywords: Medical Imaging, Pill Image, Eccentricity, Extent, Narrowness
188
Modeling Negative Affect Detector of Novice Programming Students using
Keyboard Dynamics and Mouse Behavior
Larry A. Vea12, Ma. Mercedes T. Rodrigo2
1Mapua Institute of Technology, Makati City, Philippines
([email protected])
2Ateneo de Manila University, Quezon City, Philippines
([email protected])
Abstract. We developed affective models for detecting negative affective states, particularly boredom, confusion, and frustration, among novice programming students learning C++, using keyboard dynamics and/or mouse behavior. The keystroke dynamics are
already sufficient to model negative affect detector. However, adding mouse behavior,
specifically the distance it travelled along the x-axis, slightly improved the model’s performance. The idle time and typing error are the most notable features that predominantly
influence the detection of negative affect. The idle time has the greatest influence in detecting high and fair boredom, while typing error comes before the idle time for low boredom.
Conversely, typing error has the highest influence in detecting high and fair confusion,
while idle time comes before typing error for low confusion. Though typing error is also
the primary indicator of high and fair frustrations, other features are still needed before
it is acknowledged as such. Lastly, there is a very slim chance to detect low frustration.
Keywords: Affect • model • novice programmer • keyboard dynamics • mouse behavior.
189
MOEPSO for Multi-objective Optimization
Ittikon Thammachantuek1 and Mahasak Ketcham2
Faculty of Information Technology
King Mongkut’s University of Technology North Bangkok
[email protected] , [email protected]
Abstract. This paper presents an Multi-Objective Evolutionary Particle Swarm Optimization (MOEPSO) which each iteration particles are improved by Evolutionary Algorithms. To solves Multi-Objective Optimization problem the proposed algorithm has been
compare with Multi-Objective Particle Swarm Optimization (MOPSO) using a Test Problem BNH results show that the proposed algorithm can provide more efficient solution.
Keywords: Evolutionary Particle Swarm Optimization, Multi-Objective Optimization,
Multi-Objective Evolutionary Particle Swarm Optimization
190
Multimodal Latent Feature Learning for Psyco-Physiological Stress
Modeling and Detection
Juan Lorenzo Hagad, Kenichi Fukui, and Masayuki Numao
Department of Architecture for Intelligence,
8-1 Mihogaoka, Ibaraki, Osaka, 567-0047, Japan
{hagad,fukui,numao}@ai.sanken.osaka-u.ac.jp
Abstract. Together with mobile computing and wearable medical devices, articial intelligence is poised to play an important role in the eld of mental health and stress
management. Along with a number of other advancements, using data from dierent
data modalities has been shown to be an effective way of building accurate and exible stress mod els. However, recent findings indicate that traditional machine learning
techniques may lack the ability to effectively identify salient inter-modal correlations,
especially when seemingly unrelated modalities are used concurrently. To examine this
effect, in this work we investigated the efficacy of building multimodal stress models
using a combination of psychological and physiological data. A monitoring platform
was built and unobtrusive wearable sensors were used to gather data from subjects
engaged in authentic work activities. The nal models were created by combining psychological data from stress coping proles and physiological signals from the wearable
sensors. Finally, self-annotated stress annotations to establish the ground truths used
for model training. A performance comparison was made between standard machine
learning approaches and unsupervised latent feature learning, including deep learning architectures. The results indicate that signicant improvements can be achieved
by applying deep multimodal feature learning to construct mental stress models.
Keywords: Stress Detection, Wearable Physiological Sensors, Stress Coping, Latent
Feature Learning, Multimodal Deep Learning
191
Real-time Snoring Sound Detecting U Shape Pillow System using Data
Analysis Algorithm
Patiyuth Pramkeaw1, Penpichaya Lertritchai2
, Nipaporn Klangsakulpoontawee3
Department of Media Technology,
King Mongkut’s University of Technology Thonburi, Thailand
Abstract. This paper aims to design and build snoring sound detecting u-shape pillow. Research operating includes four steps as (1) to study the problem of snoring,
(2) to analyze the related information for develop the snoring sound detecting u
shape pillow with designing the structure and sensory circuit inside the pillow. The
u-shape has been designed as a neck supporter for user. The main part of project is
the module having microphones that receive a sound of snoring. When a snoring
sound was detected, the module will command the vibrating motor to work and alert
the user to change his/her body posture. This change will help user stopping a snoring, which controlling by the C language programs, (3) to assess the quality of pillow detect snoring by five experts, (4) as result shown, the proposed pillow can detect snoring sound at 80% of accuracy based on testing with three different users.
Keywords: Snoring sound, Neck pillow, Detect snoring sound
192
The Limb Leads ECG Signal Analysis in Myocardial Infarction Patients
Anchana Muankid1 and Mahasak Ketcham2
1Department of Information Technology, Faculty of Information Technology,
King Mongkut’s University of Technology North Bangkok, Bangkok, Thailand
2Department of Information Technology management,
Faculty of Information Technology,
King Mongkut’s University of Technology North Bangkok, Bangkok, Thailand
[email protected]
Abstract. Cardiovascular disease is one of the most serious diseases in the world.
An electrocardiogram is a tool in the diagnostic of Myocardial Infarction which detects abnormal wave patterns. This paper purposes the limb leads; I, II and III electrocardiogram analysis algorithm using Wavelet transform to classify Inferior Myocardial Infarction patients. And investigate the lead which relates to inferior infarcts.
The processes in ECG signal analysis are noise elimination from the ECG signal, R
peak Detection, QRS Complex Detection and inferior Myocardial Infarction Classification. The results show that 73.33% accuracy of inferior Myocardial Infarction. Lead III and Lead II are the most relevant to inferior Myocardial Infarction.
Keywords: ECG Analysis · Wavelets Transform · Limb leads
193
Travellers’ Behaviour Analysis Based on Automatically Identified
Attributes from Travel Blog Entries
Kazuki Fujii1, Hidetsugu Nanba1, Toshiyuki Takezawa1, Aya Ishino2,
Manabu Okumura3 and Yohei Kurata4
1Hiroshima City University, 3-4-1 Ozuka-higashi,
Asaminami-ku, Hiroshima, 731-3194, Japan
{fujii, nanba, takezawa}@ls.info.hiroshima-cu.ac.jp
2Hiroshima University of Economics, 5-37-1 Gion, Asaminami-ku,
Hiroshima 731-0192, Japan
[email protected]
3Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku,
Yokohama, 226-8503, Japan, [email protected]
4Tokyo Metropolitan University, 1-1 Minamiosawa, Hachioji, Tokyo, 192-0397, Japan
[email protected]
Abstract. We propose a method to analyse travellers’ behaviour using automatically
identified travellers’ attributes, such as gender and language, from travel blog entries. We consider that travel blog entries are a useful information source for obtaining travel information, because many bloggers’ describe their travel experiences in
them. Several studies have analysed travellers’ behaviour using travel blog entries.
However, they used a small number of manually identified travellers’ (bloggers’) attributes. In our work, we identify travellers’ attributes automatically using natural language processing techniques, and conduct a large-scale travellers’ behaviour analysis.
Keywords: travel blog, behaviour analysis, travellers’ attributes
194
TSCS Monitor: Generation of Time Series Cross Section Tables from
Moodle Logs for Tracking In-Class Page Views Using Excel Macros
Konomu DOBASHI
Faculty of Modern Chinese Studies, Aichi University
4-60-6 Hiraike-cho Nakamura-ku Nagoya-shi Aichi-ken 453-8777 Japan
[email protected]
Abstract. This paper presents a method developed for viewing student access of
course materials during class time and visually capturing student engagement with
the materials. It focuses particularly on the development of TSCS Monitor, a set of
Excel macros that automatically generates Time Series Cross Section (TSCS) tables from Moodle logs in order to monitor in-class student access of course materials. The numerical data provided by the tables can be used to identify learners who
access the materials without properly following instructions or those who delay accessing the materials. It also provides data and suggestions that can be used as reference for reinforcing classroom instruction and keeping track of student engagement.
Keywords. time series, cross section, page views, student engagement, educational
data mining
195
Two Stage Travel Salesman Model of World Tourism
Surafel Luleseged Tilahun and Jean Medard T Ngnotchouye
School of Mathematics, Statistics and Computer Science,
University of KwaZulu-Natal, 3209, Pietermaritzburg, South Africa,
[email protected]
Abstract. Tourism can be defined as the commercial organization and operation of holidays and visits to places of interest. The need of visiting tourism places has increased
through time with the practice of tourism and globalization. According to UNWTO’s
long term forecast Tourism Towards 2030, international tourist arrivals worldwide are
expected to increase by 3.3% a year between 2010 and 2030 to reach 1.8 billion by 2030.
Tourism contributes for the development of a country due to its contribution to GDP, employment, exports and investment. From a tourists perspective, a tourist needs to visit as
many tourist destinations as possible with a budget constraint. In this paper, the problem
is formulated as a travel salesman problem in which a tourist want to visit each of known
tourist destination once. We assume that the cost of traveling is as a function of the distance traveled. Therefore, the problem will be minimization the travel time or distance
while visiting the intended destinations. In addition tourist destinations are aggregated
continent wise and geographically, as called clusters in this paper. Hence, the travel
salesman problem will involve another travel salesman problem for each cluster of tourism destination. Prey predator algorithm will be used to solve the proposed approach.
Keywords: Tourism, travel salesman problem, two stage optimization, prey predator
algorithm (PPA), metaheuristic
196
Variant Annotation and Clinical interpretation software for Cancer
(VARCIN): Report generating software for targeted therapy method
Pitinat Asawasutsakorn1, BenjamardMeeboon1,
Natini Jinawath2 Mingmanas Sivaraksa1
1Image, Information and Intelligence Lab,
Department of Computer Engineering, Faculty of Engineering, Mahidol University
25/25 Puttamonthon Nakhorn Pathom 73170, Thailand
1Faculty of Medicine Ramathibodi Hospital Mahidol University, 270 Rama VI Rd.,
Ratchatewi, Bangkok 10400, Thailand,
[email protected]
Abstract. Variant Annotation and Clinical Interpretation Software for Cancer (VARCIN) is a report generating software for producing an executive summary for a cancer patient. The report can assist physicians and researchers by using the result to
find a specific or recommended medicine for each specific individual with cancer,
known as “Targeted Therapy”, much more efficient. This software can shorten the
process and provide an all-in-one solution for classifying and recommending treatments for cancer patients. Moreover, this software is developed as a web-based application with a responsive feature, so this software can be used anywhere and
with any electronics device that has an access to a web browser with internet connection, improving ease-of-access thus more convenient than the traditional way.
Keywords: Variant call format, cancer, report generating, cancer diagnostic software,
targeted therapy, personalized medicine
197
Verifying Properties of Multi-agent Systems via Bounded Model Checking
IMCS, Jan Długosz University. Al. Armii Krajowej 13/15,
42-200 Cze¸stochowa, Poland.
[email protected]
Abstract. The objectives of this research are to further investigate the foundations for novel SMT and SAT-based bounded model checking (BMC) algorithms
for multi-agent systems. A major part of the research will involve the development of SMT-based BMC methods for different kinds of interpreted systems.
198
Virtual Reality System with Smartphone Application for Height Exposure
Suppanut Nateeraitaiwa1, Narit Hnoohom2
12Image, Information and Intelligence Laboratory,
Faculty of Engineering, Mahidol University, Nakorn Pathom, Thailand
Abstract. One of the treatment methods for phobias is behavioral therapy by creating
patient’s fear environment for the patient to confront that situation. Virtual reality is
one of the most interesting technologies for creating three-dimensional virtual environments. The virtual reality technology makes users feel like they are immersed in a virtual world. To accomplish creating fear environment, this paper presented a virtual reality
system with a smartphone application for height exposure. The virtual reality system
is simple, which consists of software and hardware. Users can use this system easily
in their own home. With an evaluation on 20 participants, the impact of user involvement on realism and fear of heights was explored. Paired samples t-test showed that the
user’s influence on the height of the building was significant. Moreover, the user’s influence on the realism and fear of heights when the sound is activated were significant.
Keywords: Height exposure, Virtual reality, Smartphone application, Virtual reality
glasses, Remote controller.
199
PRIMA
Abstracts
200
A Collaborative Framework for 3D Mapping using
Unmanned Aerial Vehicles
P. Doherty1, J. Kvarnström1,P. Rudol1, M. Wzorek1, G. Conte1, C. Berger1,
T. Hinzmann2, and T. Stastny2
1Dept. of Computer and Information Science, Linköping University, Sweden
2Autonomous Systems Lab, ETH Zurich, Switzerland
Abstract. This paper describes an overview of a generic framework for collaboration among humans and multiple heterogeneous robotic systems based on the use of
a formal characterization of delegation as a speech act. The system used contains a
complex set of integrated software modules that include delegation managers for each
platform, a task specification language for characterizing distributed tasks, a task
planner, a multi-agent scan trajectory generation and region partitioning module, and
a system infrastructure used to distributively instantiate any number of robotic systems and user interfaces in a collaborative team. The application focusses on 3D reconstruction in alpine environments intended to be used by alpine rescue teams. Two
complex UAV systems used in the experiments are described. A fully autonomous collaborative mission executed in the Italian Alps using the framework is also described.
A Multi Agent System for Understanding the Impact of Technology
Transfer Offices in Green-IT
Christina Herzog12, Jean-Marc Pierson1, and Laurent Lefèvre3
1IRIT, University Paul Sabatier Toulouse 3, France, {herzog,pierson}@irit.fr
2EfficIT, France
3INRIA-Lyon, ENS-Lyon, France, [email protected]
Abstract. We present a multi agent system simulating the complex interplay between the actors of innovation involved in the development of technology transfer
for Green IT. We focus on the role and the influence of technology transfer offices
on the individual objectives of each other actor (researchers, research facilities, companies). We analyse also their impact on several parameters, including sustainability.
201
Ali Baba and the Thief, Convention Emergence in Games
Xin Sun and Livio Robaldo
Faculty of Science, Technology and Communication, University of Luxembourg
{xin.sun, livio.robaldo}@uni.lu
Abstract. In this paper we propose a model that supports the emergence of norms
via multiagent learning in a social network. In our model, individual agents repeatedly interact with its neighbors over a given game, the Ali baba and Thief
game. An agent learns its strategy to play the game using the learning rule imitatethe-best. Our results show that some norms prohibiting harmful behaviors such as
“you should not rob” can emerge after repeated interactions among agents inhabited in certain social networks. Our experimental results suggest that there is critical
points of norm emergence which is decided by quotient of the initial utility and the
amount of robbery in the Ali baba and Thief game. When the quotient of the initial utility and the amount of robbery is smaller than the critical point, the probability of norm emerge is high. The probability drops dramatically as long as the quotient of the initial utility and the amount of robbery is larger than the critical point.
Analyzing Topics and Trends in the PRIMA Literature
Hoa Khanh Dam and Aditya Ghose
New South Wales 2522, Australia
{hoa,aditya}@uow.edu.au
Abstract. This study investigates the content of the literature published in the proceedings of the International Conference on Principles and Practices of MultiAgent Systems (PRIMA). Our study is based on a corpus of the 611 papers published in eighteen PRIMA proceedings from 1998 (when the conference started)
to 2015. We have developed an unsupervised topic model, using Latent Dirichlet
Allocation (LDA), over the PRIMA corpus of papers to analyze popular topics in the literature published at PRIMA in the past eighteen years. We have also
analyzed historical trends and examine the strength of each topic over time.
202
Argumentation Versus Optimization for
Supervised Acceptability Learning
Hiroyuki Kido
The University of Tokyo
7-3- 1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
[email protected]
Abstract. This paper deals with the question of how one should predict agent’s psychological opinions regarding acceptability statuses of arguments. We give a formalization of argumentation-based acceptability learning (ABAL) by introducing argument-based reasoning into supervised learning. A baseline classifier is defined based
on an optimization method of graph-based semi-supervised learning with dissimilarity
network where neighbor nodes represent arguments attacking each other, and therefore,
the optimization method adjusts them to have different acceptability statuses. A detailed comparison between ABAL instantiated with a decision tree and naive Bayes, and
the optimization method is made using each of 29 examinees’ psychological opinions
regarding acceptability statuses of 22 arguments extracted from an online discussion forum. We demonstrate that ABAL with the leave-one-out cross-validation method shows better learning performance than the optimization method in most criteria under the restricted conditions that the number of training
examples is small and a test set is used to select the best models of both methods.
203
Argumentation-Based Semantics for Logic Programs with
First-Order Formulae
Phan Minh Dung1, Tran Cao Son2, and Phan Minh Thang3
1Department of Computer Science, Asian Institute of Technology, Thailand
2Department of Computer Science, New Mexico State University, Las Cruces, NM, USA
3Department of Computer Science, Burapha University International College, Thailand
Abstract. This paper studies different semantics of logic programs with first order
formulae under the lens of argumentation framework. It defines the notion of an argumentation-based answer set and the notion of an argumentation-based well-founded model for programs with first order formulae. The main ideas underlying the new
approach lie in the notion of a proof tree supporting a conclusion given a program
and the observation that proof trees can be naturally employed as arguments in an
argumentation framework whose stable extensions capture the program’s well-justified answer semantics recently introduced in [23]. The paper shows that the proposed
approach to dealing with programs with first order formulae can be easily extended
to a generalized class of logic programs, called programs with FOL-representable atoms, that covers various types of extensions of logic programming proposed in the
literature such as weight constraint atoms, aggregates, and abstract constraint atoms.
For example, it shows that argumentation-based well-founded model is equivalent to
the well-founded model in [27] for programs with abstract constraint atoms. Finally,
the paper relates the proposed approach to others and discusses possible extensions.
204
Automatic Evacuation Management using a Multi Agent System and
Parallel Meta-heuristic Search
Leonel Aguilar, Maddegedara Lalith, Tsuyoshi Ichimura, and Muneo Hori
Earthquake Research Institute, The University of Tokyo, Bunkyo, Tokyo 113-0032
leaguilar,lalith,ichimura,[email protected]
Abstract. An automatic evacuation management system taking advantage of a multi
agent base mass evacuation simulator is proposed and prototyped. The aim of this
system is to provide a stepping stone in the direction of automated evacuation managing. The proposed system is currently capable of identifying evacuation anomalies, proposing a mitigation strategy and providing feedback for human expert evaluation and query. All the pieces although seamlessly connected are independently
developed. This allows their independent improvement and evaluation. This paper
provides an overview of the developed automatic evacuation management system and
all of its components, a demonstrative example, and discussion of its current limitations and future development direction. The demonstrative example shows increases of more than 10% in the evacuation throughput by using the proposed system.
Balancing Rationality and Utility in Logic-Based Argumentation with
Classical Logic Sentences and Belief Contraction
Ryuta Arisaka and Ken Satoh
National Institute of Informatics
Abstract. Compared to abstract argumentation theory which encapsulates the exact nature of arguments, logic-based argumentation is more specific and represents
arguments in formal logic. One significant advantage of logic-based argumentation over abstract argumentation is that it can directly benefit from logical properties such as logical consistency, promoting adherence of an argumentation framework to rational principles. On the other hand, a logical argumentation framework
based on classical logic has been also reported of its less-than- desirable utility. In
this work we show a way of enhancing utility without sacrificing so much of rationality. We propose a rational argumentation framework with just classical logic sentences and a belief contraction operation. Despite its minimalistic appearance, this
framework can characterise attack strengths, allowing us to facilitate coalition profitability and formability semantics we previously defined for abstract argumentation.
205
Can noise in behavioral models improve macro-predictions?
An empirical test
Michael Mäs1 and Dirk Helbing2
1Department of Sociology / ICS, University of Groningen
Grote Rozenstraat 31, 9712 TG Groningen, The Netherlands
2Chair of Computational Social Science, ETH Zurich
Clausiusstrasse 50, 8092 Zurich, Switzerland
Abstract. In the past decades, we have experienced a vast improvement of theories of human behavior. Nevertheless, our ability to predict the behavior of human groups, organizations, and societies has remained limited. For instance, the last financial crisis and the
Arab Spring took most scientists and experts by surprise, although theories describe with
increasing accuracy the decision-making patterns of individual stock-traders and political
activists. With two laboratory experiments, we tested whether collective phenomena can
be critically shaped by random deviations from otherwise prevalent patterns of individual
behavior. In our experiments, 96 percent of participants’ decisions were in line with a deterministic standard theory of rational decision-making. Despite this impressive microlevel accuracy, the deterministic model fails to predict the observed macro-outcomes.However, we find that a stochastic version of the same micro-theory, which
adds random deviations, largely improves macro-predictions. Our results show
that deviations can spark cascades, which fundamentally shape macro-outcomes.
The stochastic model also correctly predicted the conditions under which deviations induced system shifts.Additionally, we observed non-random deviation patterns. Our second experiment supported the hypothesis that these systematic deviations can result in fundamentally different macro-outcomes than random deviations.
206
Coalition Structure Formation using Anytime Dynamic Programming
Changder Narayan1, Dutta Animesh1, and Ghose Aditya K.2
1National Institute of Technology Durgapur, West Bengal, India
2University of Wollongong, Wollongong NSW 2522 Australia
[email protected]
Abstract. The optimal coalition structure generation is an important problem in multiagent systems that remains difficult to solve. This paper presents a novel anytime dynamic programming algorithm to compute the optimal coalition structure. The proposed
algorithm can be interrupted, and upon interruption, uses heuristic to select the largest valued coalition from each subproblem of size x and picks the rest of the unassigned agent from other subproblem of size n−x, where n is the total number of agents.
We compared the performance of our algorithm against the only existing proposal in
the literature for the optimal coalition structure problem that uses anytime dynamic
programming using 9 distinct datasets (each corresponding to a different distribution). The empirical evaluation shows that our algorithm always generates better or,
at least, as good a solution as the previous anytime dynamic programming algorithm.
Competitive VCG Redistribution Mechanism for Public Project Problem
Mingyu Guo
School of Computer Science
University of Adelaide, Australia
[email protected]
Abstract. The VCG mechanism has many nice properties, and can be applied to a
wide range of social decision problems. One problem of the VCG mechanism is that
even though it is efficient, its social welfare (agents’ total utility considering payments)
can be low due to high VCG payments. VCG redistribution mechanisms aim to resolve this by redistributing the VCG payments back to the agents. Competitive VCG
redistribution mechanisms have been found for various resource allocation settings.
However, there has been almost no success outside of the scope of allocation problems. This paper focuses on another fundamental model - the public project problem.
In Naroditskiy et al. 2012, it was conjectured that competitive VCG redistribution
mechanisms exist for the public project problem, and one competitive mechanism was
proposed for the case of three agents (unfortunately, both the mechanism and the techniques behind it do not generalize tocases with more agents). In this paper, we propose
a competitive mechanism for general numbers of agents, relying on new techniques.
207
Demand Response Integration through Agent-based Coordination of Consumers in Virtual Power Plants
Anders Clausen, Aisha Umair, Zheng Ma, and Bo Nørregaard Jørgensen
University of Southern Denmark
{ancla; aiu; zma; bnj}@mmmi.sdu.dk
Abstract. The transition towards an electricity grid based on renewable energy production induces fluctuation in electricity generation. This challenges the existing electricity
grid design, where generation is expected to follow demand for electricity. In this paper,
we propose a multi-agent based Virtual Power Plant design that is able to balance the
demand of energy-intensive, industrial loads with the supply situation in the electricity grid. The proposed Virtual Power Plant design uses a novel interagent, multi-objective, multi-issue negotiation mechanism, to coordinate the electricity demands of
industrial loads. Coordination happens in response to Demand Response events, while
considering local objectives in the industrial domain. We illustrate the applicability
of our approach on a Virtual Power Plant scenario with three simulated greenhouses. The results suggest that the proposed design is able to coordinate the electricity
demands of industrial loads, in compliance with external Demand Response events.
Dialectical Proof Procedures for Probabilistic Abstract Argumentation
Phan Minh Thang
BUUIC College, Burapha University, Thailand
Abstract. A dialectical proof procedure for computing grounded semantics of probabilistic abstract argumentation is presented based on the notion of probabilistic dispute tree.
We also present an algorithm for top-down construction of probabilistic dispute trees.
208
Distant Group Responsibility in Multi-Agent Systems
Vahid Yazdanpanah1 and Mehdi Dastani2
1University of Twente, Enschede, The Netherlands
[email protected],
2Utrecht University, Utrecht, The Netherlands
[email protected]
Abstract. In this paper, we introduce a specific form of graded group responsibility called
“distant responsibility” and provides a formal analysis for this concept in multi-agent settings. This concept of responsibility is formalized in concurrent structures based on the
power of agent groups in such structures. A group of agents is called responsible for a state
of affairs by a number of collective decision steps if there exists a strategy for the agent
group to preclude the specified state of affairs in the given number of steps. Otherwise,
the group is partially responsible based on its maximum contribution to fully responsible
groups. We argue that the notion of distant responsibility is applicable as a managerial
decision support tool for allocation of limited resources in multi-agent organizations.
209
Generalising Social Structure using Interval Type-2 Fuzzy Sets
Christopher K. Frantz1, Bastin Tony Roy Savarimuthu2, Martin K. Purvis2, and
Mariusz Nowostawski3
1College of Enterprise and Development, Otago Polytechnic, New Zealand
[email protected]
2Department of Information Science, University of Otago, New Zealand
{tony.savarimuthu, martin.purvis}@otago.ac.nz
3Faculty of Computer Science and Media Technology,
Norwegian University of Science and Technology, Norway
[email protected]
Abstract. To understand the operation of the informal social sphere in human or artificial societies, we need be able to identify their existing behavioural conventions
(institutions). This includes the contextualisation of seemingly objective facts with
subjective assessments, especially when attempting to capture their meaning in the
context of the analysed society An example for this is numeric information that abstractly expresses attributes such as wealth, but only gains meaning in its societal context. In this work we present a conceptual approach that combines clustering
techniques and Interval Type-2 Fuzzy Sets to extract structural information from aggregated subjective micro-level observations. A central objective, beyond the aggregation of information, is to facilitate the analysis on multiple levels of social organisation. We introduce the proposed mechanism and discuss its application potential.
210
Heuristics on the Data-collecting Robot Problem with immediate rewards
Zhi Xing and Jae C. Oh
Electrical Engineering and Computer Science, Syracuse University
{zxing01,jcoh}@syr.edu
Abstract. We propose the Data-collecting Robot Problem, where robots collect data as
they visit nodes in a graph, and algorithms to solve it. There are two variations of the problem: the delayed-reward problem, in which robots must travel back to the base station
to deliver the data collected and to receive rewards; and the immediate-reward problem,
in which the reward is immediately given to the robots as they visit each node. The delayed-reward problem is discussed in one of the authors’ work. This paper focuses on the
immediate-reward problem. The solution structure has a clustering step and a tour-building step. We propose Progressive Gain-aware Clustering that finds good quality solutions with efficient time complexity. Among the six proposed tour-building heuristics,
Greedy Insertion and Total-Loss algorithms perform best whendata rewards are different.
211
How, When and Where Can Spatial Segregation Induce Opinion
Polarization? Two Competing Models
Thomas Feliciani1, Andreas Flache1, Jochem Tolsma2
1ICS / Department of Sociology, University of Groningen, Groningen, The Netherlands.
2ICS / Department of Sociology, Radboud University, Nijmegen, The Netherlands.
Abstract. Increasing ethnic diversity fosters scholarly interest in how the spatial segregation of groups affects opinion polarization in a society. Despite much empirical
and theoretical research, there is little consensus in the literature on the causal link
between the spatial segregation of two groups and the emergence of opinion polarization. We contribute to the debate by investigating theoretically the conditions under
which the former fosters or hinders the latter. We focus on two processes of opinion
polarization (negative influence and persuasive argument communication) that, according to previous modeling work, can be expected to make conflicting predictions about
the relationship between segregation and opinion polarization. With a Schelling-type
ABM of residential segregation, we generate initial environments with different levels
of group segregation. Then we simulate the two processes of opinion dynamics. We
show that the negative influence model predicts segregation to hinder the emergence
of opinion polarization. On the other hand, the persuasive argument model predicts
that segregation does not substantially foster polarization. Moreover, we explore how
the spatial patterns of opinion distribution differ between the models: in particular, we
investigate the likelihood that group membership and opinion align. We show that the
alignment of group membership and opinions differs between the two opinion formation models, and that the scale at which we measure alignment plays a crucial role.
212
Individually Rational Strategy-proof Social Choice with
Wxogenous InDifference Sets
Mingyu Guo1, Yuko Sakurai2, Taiki Todo2, and Makoto Yokoo2
1The University of Adelaide
2Kyushu University
[email protected], {ysakurai, todo, yokoo}@inf.kyushu-u.ac.jp
Abstract. We consider a social choice problem where individual rationality is required.
The status quo belongs to the outcome space, and the selected alternative must be weakly better than the status quo for everybody. If the mechanism designer has no knowledge
of the alternatives, we obtain a negative result: any individually rational (IR) and strategy-proof (SP) mechanism can choose at most one alternative (besides the status quo),
regardless of the preferences. To overcome this negative result, we consider a domain
where the alternatives have a known structure, i.e., an agent is indifferent between the
status quo and a subset of the outcomes. This set is exogenously given and public information. This assumption is natural if the social choice involves the participation of
agents. For example, consider a group of people organizing a trip where participation is
voluntary. We can assume each agent is indifferent between the trip plans in which she
does not participate and the status quo (i.e., no trip). In this setting, we obtain more positive results: we develop a class of mechanisms called Approve and Choose mechanisms,
which are IR and SP, and can choose multiple alternatives as well as the status quo.
213
Modeling Organizational and Institutional Aspects in Renewable and
Natural Resources Management Context
Islem Hènane1, Sameh Hadoua25, Khaled Ghédira3, Ali Ferchichi4
1National School of Computer Sciences, Manouba University, Manouba, Tunisia
2Faculty of Economics and Management, Carthage University, Nabeul, Tunisia
3Higher Institute of Management, Tunis University, Tunis, Tunisia
4National Agronomic Institute, Carthage University, Tunisia
5Higher Colleges of Technology, Abu Dhabi, United Arab Emirates
{islemhenane, hadouaj}@yahoo.fr,
Abstract. Since 1990, there has been a striking increase in using multi-agent systems to
study renewable resources management systems. The ultimate objective is to contribute
to decisions support on resources management. The adopted strategic decisions are always joined with access to resources norms. However, the defined norms are statics and
suppose that all agents are not autonomous and always obey to the underlying norms
which do not reflect reality. In previous work, we proposed ML-MA [1], a multi-level
multi- agent architecture to support renewable resources management systems modeling.
In this work, we focus on the integration of normative aspects in our architecture. Our
approach is illustrated using “Ouled Chehida” case study from Tunisian pastoral context.
214
Offer Evaluation and Trade-off Making in Automated Negotiation
Based on Intuitionistic Fuzzy Constraints
Jieyu Zhan and Xudong Luo
Institute of Logic and Cognition, Department of Philosophy, Sun Yat-sen University,
Guangzhou, 510275, China.
Abstract. In automated negotiation, one of crucial problems is how a negotiating agent
evaluates the acceptability of an offer. Most models mainly use two kinds of evaluation
methods: (i) linear utility functions that depend on issues, and (ii) nonlinear utility functions that depend on crisp constraints. However, in real life, it is hard for human users
to input so much and so accurate information that these evaluation methods require. To
this end, this paper proposes a new approach for offer evaluation where human users are
allowed to input indeterminate information. More specifically, we propose a framework
of prioritised intuitionistic fuzzy constraint satisfaction problems for modelling agent’s
goals. Moreover, we take both satisfaction degree and dissatisfaction degree into consideration when calculating an agent’s acceptability of an offer. Finally, we discuss how
to make trade-offs via similarity measure based on intuitionistic fuzzy criteria functions.
Plan Failure Analysis: Formalization and Application in Interactive
Planning Through Natural Language Communication
Chitta Baral1, Tran Cao Son2, Michael Gelfond3 and Arindam Mitra1
1Department of Computer Science and Engineering, Arizona State University, Tempe, AZ
2Department of Computer Science, New Mexico State University, Las Cruces, NM
3Department of Computer Science, Texas Tech University, Lubbock, TX
Abstract. While most robots in human robot interaction scenarios take instructions
from humans, the ideal would be that humans and robots collaborate with each other. The Defense Advanced Research Projects Agency Communicating with Computer
program proposes the collaborative blocks world scenario as a testbed for this. This
scenario requires the human and the computer to communicate through natural language to build structures ou of toy blocks. To formulate and address this, we identify two main tasks. The first task, called the plan failure analysis, demands the robot to analyze the feasibility of a task and to determine the reasons(s) in case the
task is not doable. The second task focuses on the ability of the robot to understand
communications via natural language. We discuss potential solutions to both problems and present prototypical architecture for the integration of planning failure
analysis and natural language communication into an intelligent agent architecture.
215
Resistance to Corruption of General Strategic Argumentation
Michael J. Maher
School of Engineering and Information Technology
University of New South Wales, Canberra
ACT 2600, Australia
E-mail: [email protected]
Abstract. [16, 18] introduced a model of corruption within strategic argumentation,
and showed tha some forms of strategic argumentation are resistant to two forms of
corruption: collusion and espionage. Such a model provides a (limited) basis on which
to trust agents acting on our behalf. However, that work only addressed the grounded and stable argumentation semantics. Here we extend this work to several other
well-motivated semantics. We must consider a greater number of strategic aims that
players may have, as well as the greater variety of semantics. We establish the complexity of several computational problems related to corruption in strategic argumentation, for the aims and semantics we study. From these results we identify that strategic
argumentation under the aims and semantics we study is resistant to espionage. Resistance to collusion varies according to the player’s aim and the argumentation semantics, and we present a complete picture for the aims and semantics we address.
216
Revenue Maximizing Markets for Zero-Day Exploits
Mingyu Guo1, Hideaki Hata2, and Ali Babar1
1School of Computer Science
University of Adelaide, Australia
{mingyu.guo, ali.babar}@adelaide.edu.au
2Graduate School of Information Science
Nara Institute of Science Technology, Japan
[email protected]
Abstract. Markets for zero-day exploits (software vulnerabilities unknown to the
vendor) have a long history and a growing popularity. We study these markets from
a revenue-maximizing mechanism design perspective. We first propose a theoretical
model for zero-day exploits markets. In our model, one exploit is being sold to multiple buyers. There are two kinds of buyers, which we call the defenders and the offenders. The defenders are buyers who buy vulnerabilities in order to fix them (e.g.,
software vendors). The offenders, on the other hand, are buyers who intend to utilize
the exploits (e.g., national security agencies and police). Our model is more than a
single-item auction. First, an exploit is a piece of information, so one exploit can be
sold to multiple buyers. Second, buyers have externalities. If one defender wins, then
the exploit becomes worthless to the offenders. Third, if we disclose the details of the
exploit to the buyers before the auction, then they may leave with the information
without paying. On the other hand, if we do not disclose the details, then it is difficult for the buyers to come up with their private valuations. Considering the above,
our proposed mechanism discloses the details of the exploit to all offenders before the
auction. The offenders then pay to delay the exploit being disclosed to the defenders.
217
Semantic Reasoning with Uncertain information from Unreliable Sources
Murat Şensoy1, Lance Kaplan2, and Geeth de Mel3
1Department of Computer Science, Ozyegin University, Istanbul, TR.
2US Army Research Lab, Adelphi, MD 20783, USA.
3IBM T. J. Watson Research Center, Hawthorne, NY, USA.
Abstract. Intelligent software agents may significantly benefit from semantic reasoning.
However, existing semantic reasoners are based on Description Logics, which cannot
handle vague, incomplete, and unreliable knowledge. In this paper, we propose SDL-Lite which extends DL-Lite R with subjective opinions to represent uncertainty in knowledge. We directly incorporate trust into the reasoning so that the inconsistencies in the
knowledge can be resolved based on trust evidence analysis. Therefore, the proposed
logic can handle uncertain information from unreliable sources. We demonstrate how
SDL-Lite can be used for semantic fusion of uncertain information from unreliable sources and show that SDL- Lite reasoner can estimate the ground truth with a minimal error.
Sequence Semantics for Normative Agents
Guido Governatori1, Francesco Olivieri2, Erica Calardo3,
Antonino Rotolo3, and Matteo Cristani2
1Data61, CSIRO, Australia
2University of Verona, Italy
3University of Bologna, Italy
Abstract. We proposed a novel framework for the representation of goals and other
mental-like attitudes in terms of degree of expected outcomes, where an outcome is an
order of possible alternatives. The sequences of alternatives is modelled by a non-classical (substructural) operator. In this paper we provide a modal logic based axiomatisation of the intuition they propose, and we discuss some variants (in particular for
the notion of social intention, intentions that are compliant with norms). Given that
the outcome operator is substructural, we first propose a novel sequence semantics (a
generalisation of possible world semantics) to model the outcome operator, and we
prove that the axiomatisation is sound and complete with respect to the new semantics.
218
Spread of Cooperation in Complex Agent Networks Based on
Expectation of Cooperation
Ryosuke Shibusawa, Tomoaki Otsuka, and Toshiharu Sugawara
Department of Computer Science and Communications Engineering,
Waseda University, Tokyo 1698555, Japan
Abstract. This paper proposes a behavioral strategy called expectation of cooperation with which cooperation in the prisoner’s dilemma game spreads over agent networks by incorporating Q-learning. Recent advances in computer and communication technologies enable intelligent agents to operate in small and handy computers
such as mobile PCs, tablet computers, and smart phones as delegates of their owners. Because the interaction o these agents is associated with social links in the real
world, social behavior is to some degree required to avoid conflicts, competition, and
unfairness that may lead to further inefficiency in the agent society. The proposed
strategy is simple and easy to implement but nevertheless can spread over and maintain cooperation in agent networks under certain conditions. We conducted a number of experiments to clarify these conditions, and the results indicate that cooperation spread and was maintained with the proposed strategy in a variety of networks.
219
Towards Better Crisis Management in Support Services Organizations
using Fine Grained Agent Based Simulation
Vivek Balaraman, Harshal Hayatnagarkar, Meghendra Singh, Mayuri Duggirala
Human Centric Systems, TCS Research,
54-B, Hadapsar Industrial Estate, Pune, India.
{vivek.balaraman, h.hayatnagarkar2, meghendra.singh,
mayuri.duggirala}@tcs.com
Abstract. Critical support service operations have to run 24 x 7 and 365 days a year.
Support operations therefore do contingency planning to continue operations during a
crisis. In this paper we explore the use of fine-grained agent-based simulation models,
which factor in human-behavioral dimensions such as stress, as a means to do better
people planning for such situations. We believe the use of this approach may allow support operations managers to do more nuanced planning leading to higher resilience, and
quicker return to normalcy. We model a prototypical support operation, which runs into
different crisis severity levels, and show for each case, a reasonable size of the crisis team
that would be required. We identify two contributions in this paper: First, emergency
planning using agent based simulations have mostly focused, naturally, on societal communities such as urban populations. There has not been much attention paid to study
crisis responses within support services organizations and our work is an attempt to address this deficit. Second, our use of grounded behavioral elements in our agent models
allows us to build complex human behavior into the agents without sacrificing validity.
Verifying Real-Time Properties of Multi-Agent Systems via SMT-based
Bounded Model Checking
Agnieszka M. Zbrzezny and Andrzej Zbrzezny
IMCS, Jan Długosz University. Al. Armii Krajowej 13/15, 42-200 Czȩstochowa,
Poland.
{agnieszka.zbrzezny, a.zbrzezny}@ajd.czest.pl
Abstract. We present a satisfiability modulo theories based bounded model checking (SMT-based BMC) method for timed interpreted systems (TIS) and for properties expressible in the existential fragment of a Real-Time Computation Tree
Logic with epistemic components (RTECTLK). We implemented the standard
BMC algorithm and evaluated it for two multi-agent systems: a timed train controller system and a timed generic pipeline paradigm. Weused the Z3 solver.
220
PRIMA Student
Abstracts
221
A Kernelization approach for Anytime Coalition Structure Generation
using Knuth X algorithm
Narayan Changder, Animesh Dutta, and Aditya Ghose
National Institute of Technology, Durgapur
Abstract. Determining the optimal coalition structure is an interesting problem in
multi-agent system. The main problem is, after a group of agents comes together, how
the agents can be partitioned such that social payoff is maximized? This work presents the application of Knuth X algorithm in coalition structure generation problem.
Freight Train Scheduling Problem
Samriddhi Sarkar and Animesh Dutta
National Institute of Technology Durgapur, India
Abstract. In most of the countries, passenger and freight trains share the same infrastructure. However, freight trains are run without any fixed schedule, they are operated
as the requirements arise. This paper adopts Timed Colored Petri Nets theory to model
the railway system. Proposed system will automatically detect the possible conflicts
amongst the freight request and passenger trains’ schedule using various TCPN properties. After that system will try to find out a feasible schedule for the freight train taking
into account the safety rules set by railways, that too, without any human intervention.
222
Multi-Agent Based Scalable and Context-Aware Middleware for
Typical IoT Scenarios
Bikash Choudhury, Subhrabrata Choudhury, and Animesh Dutta
Abstract. The objective of the work is to design a multi-agent based distributed
middleware forservice oriented computing (SOC) in mobile pervasive environment
that improves the QoSof the service consumers with minimal resources usage in the
system. From another angle it can also be viewed as a system that enables on-demand self-organised Fog creation in the vicinity of the service consumer that reduces the traffic load at core network and the resource requirement of the cloud servers.
Real-time Collision Handling in Railway Network:
An Agent-based Approach
Poulami Dalapati and Animesh Dutta
National Institute of Technology durgapur, India
Abstract. This research work addresses the issues concerning collision detection and
resolution in a complex railway network. The proposed approach tries to resolve conflicts
during runtime in distributed manner satisfying real time constraints using cooperative autonomous agents. The objective function of this constraint satisfaction problem considers
early detection of as well as minimization of total number of conflicts in a time interval.
223
Role of Facial Emotion in Social Correlation
Pankaj Mishra, Rafik Hadfi, and Takayuki Ito
Abstract. Diffusion of information can be found in any social network, either it be
a small network of co-workers or large as social networking applications like Facebook, Twitter, LinkedIn, etc. In this paper, we propose a system to analyse the correlation amongst the group of people, by analysing the emotion diffusion of all the
participants in the considered network or group. Also, we implement our system
in a multi-agent paradigm to maintain the robustness and scalability of the algorithm. Finally, we test our algorithm with the scripted discussion, of which ground
truth correlation is known. Such knowledge of the influences among the participants of the network, could be used to find the influential individual; further
can find many applications such in consensus, negotiation, viral marketing, etc.
Security and Access Control in Multi-Agent Satellite Systems
Pratik Sinha and Animesh Dutta
Abstract. This research work aims to address the issues related to security and access control in multi-agent satellite systems. The communication among agents
and ground stations are prone to interception. Also, agents should perform tasks
given by authorized ground stations and agents only. However, ground stations
must not be involved repetitively for access control. The proposed work aims
to formally model the problem and develop appropriate algorithmic solutions.
224
Selected Methods of Model Checking using SAT and SMT-solvers
Agnieszka Zbrzezny
Institute Of Mathematics and Computer Science,
Jan Dlugosz University in Czestochowa
Abstract. The objectives of this research are to further investigate the foundations for novel SMTand SAT-based bounded model checking (BMC) algorithms for real-time and multi-agent systems. A major part of the research will
involve the development of SMT-based BMC methods for different kinds of
Kripke structures, interpreted systems for different kinds of temporal languages.
225
PKAW
Abstracts
226
A New Hybrid Rough Set and Soft Set Parameter Reduction Method for
Spam E-mail Classification Task
Masurah Mohamad and Ali Selamat
Abstract. Internet users are always being attacked by spam messages, especially spam
e-mails. Due to this issue, researchers had done many research works to find alternatives against the spam attacks. Different approaches, software and methods had been
proposed in order to protect the Internet users from spam. This proposed work was
inspired by the rough set theory, which was proven effective in handling uncertainties and large data set and also by the soft set theory which is a new emerging parameter reduction method that could overcome the limitation of rough set and fuzzy
set theories in dealing with an uncertainty problem. The objective of this work was to
propose a new hybrid parameter reduction method which could solve the uncertainty
problem and inefficiency of parameterization tool issues which were used in the spam
e-mail classification process. The experimental work had returned significant results
which proved that the hybrid rough set and soft set parameter reduction method can
be applied in the spam e-mail classification process that helps the classifier to classify
spam e-mails effectively. As a recommendation, enhancement works on the functionality of this hybrid method shall be considered in different application fields, especially for the fields dealing with uncertainties problem and high dimension of data set.
227
Abbreviation Identification in Clinical Notes with Level-wise Feature
Engineering and Supervised Learning
Vo Thi Ngoc Chau, Cao Hoang Tru, Ho Tu Bao
Abstract. Nowadays, electronic medical records get more popular and significant in
medical, biomedical, and healthcare research activities. Their popularity and significance lead to a growing need for sharing and utilizing them from the outside. However,
explicit noises in the shared records might hinder users in their efforts to understand and
consume the records. One kind of explicit noises that has a strong impact on the readability of the records is a set of abbreviations written in free text in the records because
of writing-time saving and record simplification. Therefore, automatically identifying
abbreviations and replacing them with their correct long forms are necessary for enhancing their readability and further their sharability. In this paper, our work concentrates on abbreviation identification to lay the foundations for de-noising clinical text
with abbreviation resolution. Our proposed solution to abbreviation identification is
general, practical, simple but effective with level-wise feature engineering and a supervised learning mechanism. We do level-wise feature engineering to characterize each
token that is either an abbreviation or a non-abbreviation at the token, sentence, and
note levels to formulate a comprehensive vector representation in a vector space. After
that, many open options can be made to build an abbreviation identifier in a supervised
learning mechanism and the resulting identifier can be used for automatic abbreviation
identification in clinical text of the electronic medical records. Experimental results
on various real clinical note types have confirmed the effectiveness of our solution
with high accuracy, precision, recall, and F-measure for abbreviation identification.
Acquiring Seasonal/Agricultural Knowledge from Social Media
Hiroshi Uehara and Kenichi Yoshida
Abstract. Agricultural knowledge depends on seasonally changing conditions
such as climate, harmful insects, etc. In this respect, farmers tend to be interested in seasonal knowledge rather than the static principle. To acquire such agricultural knowledge, we propose a method to acquire seasonal knowledge from ongoing posts in the social media. The experimental results shows that the agricultural
knowledge can be extracted in the form of chained structures, each of which denotes
a set of seasonal knowledge. We also developed a prototype of dialogue robot that
provides agricultural knowledge based on the chained structure database. The characteristics of the robot is its ability to reply with seasonally changing knowledge.
228
Amalgamating Social Media Data and Movie Recommendation
Maria R. Lee, Tsung Teng Chen, Ying Shun Cai
Abstract. Recommender systems (RSs) have become very common recently. However, RS techniques need large amounts of user and product data, which hinders RS
usage for businesses with insufficient data. The RS cold-start problem may be mitigated by leveraging external data sources. We demonstrate the feasibility of solving
the cold-start problem by implementing a hybrid RS that integrates the Facebook Fan
Page data and the genre-classifications data from Yahoo! Movies. Our study amalgamates social media data and machine learning to build a hybrid-filtering RS. We also
compared our system with three existing movie RSs?those used by Netflix, YouTube,
and Amazon. Within the framework of a hybrid-filtering RS, content-based filtering
was used to extract data from Yahoo! Movies and Facebook Fan Pages. The proposed
RS overcame the cold-start problem and achieved a satisfactory level of accuracy.
Building a Mental Health Knowledge Model to Facilitate
Decision Support
Bo Hu, Boris Villazon Terrazas
Abstract. Medical research produces a vast amount of data everyday through for
instance high throughput preclinical and clinical tools. Exploiting such a source
of knowledge, as well as discovering patterns and relations buried within, can offer great help to clinical professionals in high quality health care services. There is
a growing reliance on advanced computing technologies to help make sense and
comprehend such data. In this paper, we describe the application of Word2Vec to
facilitate knowledge discovery from very-large public unstructured text corpora
(worked with PubMed thus far, but can easily incorporate others). Benefit from unsupervised word embedding, we experiment how new knowledge can stem from
peer-reviewed medical publications and cross-reference such knowledge with established one to understand the advantages and disadvantages of popular deep-learning
based approaches to knowledge acquisition. We also developed a proof-of-concept
computer system to exploit such knowledge in a medical recommendation system.
229
Building a Process Description Repository with Knowledge Acquisition
Diyin Zhou, Hye-Young Paik, Seung Hwan Ryu, John Shepherd, Paul Compton
Abstract. Although there is an abundance of how-to guides online, systematically
utilising the collective knowledge represented in such guides has been limited. This
is primarily due to how-to guides (effectively, informal process descriptions) being
expressed in natural language, which complicates the process of extracting actions
and data. This paper describes the use of Ripple-Down Rules (RDR) over the Stanford NLP toolkit to improve the extraction of actions and data from process descriptions in text documents. Using RDR, we can incrementally and rapidly build rules
to refine the performance of the underlying extraction system. Although RDR has
been widely applied, it has not so far been used with NLP phrase structure representations. We show, through implementation and evaluation, how the use of action-data extraction rules and knowledge acquisition in RDR is both feasible and effective.
Building a Working Alliance with a Knowledge Based System through an
Embodied Conversational Agent
Deborah Richards, Patrina Caldwell
Abstract. Knowledge is only useful if the intended user is able to utilize the system.
In early knowledge based systems, known as expert systems, the user interface provided the means through which knowledge was acquired and accessed. Today it is possible to interact with the knowledge of an expert through a humanlike interface, in
the form of an embodied conversational agent (ECA). Through familiar conversational-style interaction, the ECA can obtain the state of the user and provide a recommendation overcoming health literacy barriers and, depending on the nature of the dialogue,
build a working alliance with the human that will encourage adherence to the advice.
In this paper we describe the eADVICE system in the domain of paediatric incontinence that aims to improve adherence through the use of an ECA as the interface to
the domain knowledge. Results of an initial pilot are provided showing that those who
used the ECA achieved improved health outcomes and found the experience positive.
230
Combining Feature Selection with Decision Tree Criteria and
Neural Network for Corporate Value Classification
Ratna Hidayati; Katsutoshi Kanamori; Ling Feng; Hayato Ohwada
Abstract. This study aims to classify corporate values among Japanese companies
based on their corporate social responsibility (CSR) performances. Since there are
many attributes in CSR, feature selection with decision tree criteria is used to select
the attributes that can classify corporate values. The feature selection found that 41%
of 37 total attributes, or only 15 attributes, are needed to classify corporate values.
The accuracy of building the tree used to find the 15 attributes is low. To increase the
accuracy, the attributes are trained in a neural network. The accuracy of the decision
tree is 0.7, and the accuracy of the neural for training the 15 attributes increased to
0.75. To sum up, this study found, companies with higher corporate values seek to
enhance their CSR activities or to empower secondary stakeholders. In contrast, companies with low corporate values still focus their CSR activities on primary stakeholders.
Competition Detection from Online News
Zhong-Yong Chen and Chien Chin Chen
Abstract. In this paper, we define a novel problem named competed intention identification of online news. We propose new features to represent the competed intention of the
documents. The support vector machine (SVM) is employed to adopt our features to identify the competed intention in the news article. Experimental results demonstrate that the
features we designed are effective for identifying the documents with competed intention.
231
Enhanced Rules Application Order to Stem Affixation, Reduplication and
Compounding Words in Malay Texts
Mohamad Nizam Kassim, Mohd Aizaini Maarof,
Anazida Zainal, Amirudin Abdul Wahab
Abstract. Word stemmer is an automated program to remove affixes, clitics and particles
from derived words based on morphological structures of specific natural languages. It
has been widely used for text preprocessing in many artificial intelligence applications.
Furthermore, the performance of word stemmer to correctly stem derived words has an
influence to the performance of information retrieval, text mining and text categorization
applications. Despite of various stemming approaches were proposed in the past research,
the existing word stemmers for Malay language still suffer from stemming errors. Moreover, the existing word stemmers partially consider morphological structures of Malay
language in which only focused on affixation words instead of affixation, reduplication
and compounding words, simultaneously. Therefore, this paper proposes an enhanced
word stemmer using rule-based affixes removal and dictionary lookup methods called
enhanced rule application that is able to stem affixation, reduplication and compounding
words and at the same time, is able to address possible stemming errors. This paper also
examines possible root causes of affixation, reduplication and compounding stemming
errors that could happen during word stemming process. The experimental results indicate that the proposed word stemmer is able to stem affixation, reduplication and compounding words with better stemming accuracy by using enhanced rule application order.
Finding Reliable Source for Event Detection Using Evolutionary Method
Raushan Ara Dilruba and Mahmuda Naznin
Abstract. Participatory sensing is a phenomenon where participants use mobile
phones or social media and feed data to detect an event. Since, data gathering is open
to many participants, one of the major challenges of this type of networks is to identify truthfulness of the reported observations. Finding the reliable sources is a challenging task since the node or participant’s reliability is unknown or even the probability of the reported event to be true is also unknown. In our paper, we study this
challenge and observe that applying evolutionary method, we can identify reliable
source nodes. We call our approach Population Based Reliability Estimation. We validate our claim by experimental results. We also compare our method with another
widely used method. From experiments we find that our approach is more efficient.
232
Improving Motivation in Survey Participation by Question Reordering
Rohit Kumar Singh, Vorapong Suppakitpaisarn, and Ake Osothongs
Abstract. We organized an experiment to show that survey participants take part more
when the questionnaires started with less aggressive questions. In our earlier work, we
used Bayesian probability and graph algorithms to find relative values of each personal
attribute. Using that valuation, we created two sets of the questionnaire each differs in
question order and ask 33 personal attributes from participants. The first set of the questionnaire ordered questions from personal attributes with high valuations such as passport
number, driving license number, last name, and monthly income to personal attributes
with low valuations such as nationality, gender and office country. On the other hand, the
second set of questionnaire ordered from those with low valuations to those with higher
valuations. As a result, the number of participants who received the second set of the questionnaire and agrees to submit some information is 71.42% more than those who received
the first set of the questionnaire. Moreover, the second set of participants spends much
less time in filling the questionnaire, but provides 1.78% more information on average.
Integrating Symbols and Signals based on Stream Reasoning and ROS
Takeshi Morita, Yu Sugawara, Ryota Nishimura, and Takahira Yamaguchi
Abstract. We have developed PRactical INTElligent aPplicationS (PRINTEPS)
which is a total intelligent application development platform. This paper introduces an application of PRINTEPS for detecting events by using stream reasoning and Robot Operating System (ROS), and for integrating image sensing with
knowledge proces ing. Based on this platform, we demonstrate that the behaviors of a robot in a robot cafe can be modified by changing the applicable rule sets.
233
Knowledge Acquisition for Learning Analytics:
Comparing teacher-derived, algorithm-derived,
and Hybrid Models in the Moodle Engagement Analytics Plugin
Danny Y.T. Liu, Deborah Richards, Phillip Dawson,
Jean-Christophe Froissard, Amara Atif
Abstract. One of the promises of big data in higher education (learning analytics)
is being able to accurately identify and assist students who may not be engaging as
expected. These expectations, distilled into parameters for learning analytics tools,
can be determined by human teacher experts or by algorithms themselves. However, there has been little work done to compare the power of knowledge models acquired from teachers and from algorithms. In the context of an open source learning
analytics tool, the Moodle Engagement Analytics Plugin, we examined the ability of
teacher-derived models to accurately predict student engagement and performance,
compared to models derived from algorithms, as well as hybrid models. Our preliminary findings, reported here, provided evidence for the fallibility and strength
of teacher- and algorithm-derived models, respectively, and highlighted the benefits
of a hybrid approach to model- and knowledge-generation for learning analytics. A
human in the loop solution is therefore suggested as a possible optimal approach.
Learning under Data Shift for Domain Adaptation: A Model-based
Co-clustering Transfer Learning Solution
Santosh Kumar, Xiaoying Gao, and Ian Welch
Abstract. Data shifting in machine learning problems violates the common assumption that the training and testing samples should be drawn from the same distribution.
Most of the algorithms which provide the solution for data shifting problems first try
to evaluate the distributions and then reweight samples based on their distributions.
Due to the difficulty of evaluating a precise distribution, conventional methods cannot achieve good classification performance. In this paper, we introduce two types of
data-shift problems and propose a modelbased co-clustering transfer learning based
solution which consistently deals with both scenarios of data shift. Experimental results demonstrate that our proposed method achieves better generalization and running efficiency compared to traditional methods under data or covariate shift setting.
234
Predicting the scale of trending topic diffusion among online communities
Dohyeong Kim, Soyeon Caren Han, Sungyoung Lee, and Byeong Ho Kang
Abstract. Online trending topics represent the most popular topics among users in certain
online community, such as a country community. Trending topics in one community are
different from others since the users in the community may discuss different topics from
other communities. Surprisingly, almost 90% of trending topics are diffused among multiple online communities, so it shows people’s interests in a certain community can be
shared to others’ in another community. The aim of this research is to predict the scale of
trending topic diffusion among different online communities. The scale of diffusion represents the number of online communities that a trending topic diffuses. We proposed a
diffusion scale prediction model for trending topics with the following four features, including community innovation feature, context feature, topic feature, and rank feature.
We examined the proposed model with four different machine learning in predicting
the scale of diffusion in Twitter Trending Topics among 8 English-speaking countries.
Our model achieved the highest prediction accuracy (80.80%) with C4.5 decision tree.
Quality of Thai to English Machine Translation
Seamus Lyons
Abstract. This paper presents the experimental results of several approaches to
machine translation evaluation to determine the quality of Thai to English translation. We compare automatic metrics and human-based evaluation that includes
error classification, reading comprehension and analysis from a professional translator. The research compares translation systems that are available to end users in
Thailand to provide an understanding of the quality of translation in general use.
Both the rate of 47.2% error words per text and the BLEU score of 0.21 indicate
the difficulty of Thai to English translation. Despite a high error rate for the translations, users were able to successfully answer about 60% of the questions using
the output of the machine translation systems in the reading comprehension tests.
235
Robust Modified ABC Variant (JA-ABC5b) for Solving Economic
Environmental Dispatch (EED)
Noorazliza Sulaiman, Junita Mohamad-Saleh, and Abdul Ghani Abro
Abstract. Artificial bee colony (ABC) algorithm has been widely used in solving various optimization problems due to its simplicity and flexibility besides showing outrageous results in comparison to other optimization algorithms. Nevertheless, ABC has
been found to suffer from few limitations such as slow convergence rates and premature convergence tendency. With the motivation to overcome the problem, this work
proposes a modified ABC variant referred to as JA-ABC5b with the aim of robust and
faster convergence. The proposed ABC variant has been compared with the standard
ABC and other existing ABC variants on 27 benchmarks functions and to solve economic environmental dispatch (EED) problem. The results have shown that JAABC5b has the best performance in comparison to the standard ABC and selected existing
ABC variants in terms of convergence speed and global optimum achievement besides
exhibits robust performance in solving complex real-world optimization problems.
Specialized Review Selection Using Topic models
Anh Duc Nguyen, Nan Tian, Yue Xu, and Yuefeng Li
Abstract. Online reviews and comments about a product or service are an invaluable
source of information for users to assist them in making purchase decisions. In recent
years, the research in review selection has attracted considerable attention. Many of the
existing works attempted to identify a number of statistical features related to review
text such as word count (Mudambi & Schuff, 2010) and hidden relations between these
features and review quality by using supervised learning methods such as classification
techniques. However, one significant drawback of these works is that they do not take the
review content into consideration. A recent work has been proposed to find specialized
reviews that focus on a specific feature based on similar words to the feature (Long et al.,
2014). In this paper, we propose a topic model based method which selects reviews by
considering both similar words and related words from a topic model such as LDA model. The conducted experiment has proven that those related words generated from LDA
have a great contribution to the task of finding helpful reviews on a specified feature.
236
Stable Matching in Structured Networks
Ying Ling, Tao Wan and Zengchang Qin
Abstract. Stable matching studies how to pair members of two sets with the objective to
achieve a matching that satisfies all participating agents based on their preferences. In this
research, we consider the case of matching in a social network where agents are not fully connected. We propose the concept of D-neighbourhood associated with connective
costs to investigate the matching quality in four types of well-used networks. A matching
algorithm is proposed based on the classical Gale-Shapley algorithm under constraints of
network topology. Through experimental studies, we find that the matching outcomes in
scale-free networks yield the best average utility with least connective costs comparing
to other structured networks. This research provides insights for understanding matching
behavior in social networks like marriage, trade, partnership, online social and job search.
Workflow interpretation via Social Networks
Eui Dong Kim and Dr. Peter Busch
Abstract. We sought to determine how people worked in practice, how management
saw they worked and examine ‘gaps’ between these two ‘views’. In order to see potential differences, we examined workflow management through interviews with
managers and a questionnaire with employees. The results were analysed through
Petri Nets in a simplified form. The second unit of analysis was examining relationships between employees and therefore their knowledge flows using Social Network
Analysis to illustrate work patterns staff had with one another. Through overlaying
the two we gained some understanding of matches and mismatches. The study took
place in three IT units of one organisation - an Australian university. The outcomes
of our study comprise potential recommendations for improving work efficacy, such
as re-organising work practices, or potentially changing who works with whom.
237
Fuzzy C-Means Classification of Electroencephalography (EEG) Waves
for Robotic System Time Events and Control
Ebrahim A. Mattar1 and Hessa J. Al-Junaid2
1College of Engineering, University of Bahrain, P.O. Box 32038, Kingdom of Bahrain
[email protected]
2College of Information Technology, University of Bahrain, P.O. Box 32038,
Kingdom of Bahrain
[email protected]
Abstract. The brain is the most complicated part of any intelligent life forms, it controls
almost everything in the body. For controlling a system via a brain, there must be communication. After years of medical research, it was established that, brains communicate with
origins via massive set of electrical signals (Electroencephalography (EEG)) that are periodically sent. Such (EEG) waves are sent from neurons in the brain to the ones in the spinal
cord and end in nerve endings in the origins. The neurons are connected together, in addition
to nerves using synapses. This is the basic structure of the nervous system that is in charge
of communications between the brain and the rest of the body. The brain signals are basically
electrical currents. Recently, also EEG-Based robotics systems control has received a fundamental attention by researchers world-wide. This is due to the advancement of related propping and sensing technologies for non-invective propping of the complicated, and coupled
neural brains waves, in addition to the beneficial use and background gained for such BCI
technologies. In particular, moving a robotic system via human brainwaves, has received a
focal attention due to the advantages resulting from this technology for robotics and rehabilitations applications. One of the major issue that is needed to be looked into is related to the
DECODING and PATTERNS of the resulting EEG waves for motorizing applications. There
are a number of decoding methods and techniques, however, we shall will look into FUZZY
C-MEANS related decoding. In this respect, the aim of this paper is to use fuzzy c-mean
clustering for EEG waves, in such a way to control a robotic system using a set of detected
EEG brainwaves. The EEG waves will be analysed based on fuzzy c-mean routines, in such
a way to detect and define events that are of meaningful to the robotic system under control.
The EEG decoding will involve a number of procedures, as this will lead to the classification
routines. In addition, a fuzzy c-means classification technique will used, as this will help in
building a fuzzy based rule system. For accomplishing this study, we shall rely on ready
detected and measured EEG brainwaves (from well-known International Laboratories), to
ensure the stability, accuracy, and the data. Subsequently, the technique will also be used for
allowing the EEG waves control a robotic system. The system under control will be a robotic
arm system, as such EEG will help in motorising the robotic arm in small scale moving space.
Keywords: Fuzzy C-Mean, Clustering, Brain-controlled robot, Brain-Machine Interface (BMI), Brain-Computer, Interface (BCI), Human-Machine integration control,
Electroencephalogram (EEG).
238
Proceedings Online Access (Springer)
• PRICAI 2016: Trends in Artificial Intelligence
http://link.springer.com/book/10.1007/978-3-319-42911-3
• PRIMA 2016: Princiles and Practice of Multi-Agent Systems
• Knowledge Management and Acquisition for Intelligent Systems (PKAW)
239
Guides
Currency (Thai Baht, Notes)
Foreign Currencies
US Dollar
Euro
British Pound Sterling
Japanese Yen: 100
Singapore Dollar
Hong Kong Dollar
Korean Won
Swiss Franc
Australian Dollar
Malaysian Ringgit
Sounth African Rand
Swedish Krona
Canadian Dollar
New Zealand Dollar
Indian Rupee
Chinese Yuan
Philippine Peso
Taiwan Dollar
Indonesian Rupiah
Russia Ruple
Vietnamese Dong
USD
EUR
GBP
JPY
SGD
HKD
KRW
CHF
AUD
MYR
ZAR
SEK
CAD
NZD
INR
CHY
PHP
TWD
IDR
RUB
VND
Selling Rates
35.25
39.40625
47.19125
35.265
26.3875
4.67375
0.0365
36.46875
27.73375
9.12
3.9
4.14
26.95375
25.9775
0.596
5.528
0.85
1.26
0.0043
0.86
0.0018
Buying Rates
34.5
37.95875
44.31625
33.48
25.29875
4.35845
0.0275
34.6725
25.70375
8.3
1.9
3.885
25.64875
24.28125
0.4
4.79
0.53
0.86
0.0019
0.38
0.0011
Currency Exchange Rates from Siam Commercial Bank (9th August 2016)
240
About Phuket
What comes into the mind of travellers when we talk about sea, sun and sand?
Phuket must definitely be one of the answers. A number of exciting activities can be
found on this island. In the early days of regional maritime trade, the cape of Phuket
was locally referred to as Jung Ceylon, while locals called it
Thalang, which evolved to be the name of the main town to
the north of the island. As the perfect stopover sheltering traders from monsoons, Jung Ceylon welcomed merchants from
India, Persia, Arabia, Burma, China, and also Siam. During
the 16th century, the island was a popular trading port for tin.
In 1785, Thalang town was surrounded by Burmese troops
who invaded the coastal area. It was under the leadership of
Chan, the widow of the governor, and her sister, Muk, who
united the local residents and successfully fought and drove the invaders out of Phuket.
It took over 30 days for the defending troops of Phuket, under the command of Chan
and Muk, to claim their victory. As a result of such heroic deeds, noble titles were
granted to Chan and Muk as Thao Thep Kasattri and Thao Sri Soonthorn, respectively. They are still highly respected by Phuket
residents even today. When the city was in
a peaceful state, the development of mining
was so unprecedented. Chinese businessmen and miners later migrated to Phuket
and soon enjoy thriving wealth. The island’s
long history has shaped the Phuket of the
present with its diverse ethnic groups, culture, architectural influence, and fine cuisine. These attributes have made Phuket a complete tourist destination that offers a
lot more beyond its natural heritage of sea, sand, forest, and world-renowned diving
241
sites. Sino-Portuguese architecture casts its spell delighting travellers to the city, while
Phuket style of hospitality has never failed to impress visitors from all walks of life.
As the most popular island destination in Thailand, Phuket has numerous options for traveling to the island and getting around once you arrive. Both domestic
and international airlines service Phuket Airport with direct flights from numerous destinations in Thailand and around Asia. Once on the island, the size of Phuket
makes a rental car arguably the best option, though there are various modes of
transportation if you do not wish to drive. Source: http://www.tourismthailand.org
Getting Around in Phuket
Car Rental Service: There are numerous car rental services on Phuket. Cars or jeeps
can be rented at the airport, in Phuket Town, and at most of the more popular beaches. Be
aware that only Commercial First Class Insurance provides full coverage on rental cars
(as opposed to limited personal or third party only insurance). Most international car
rental agencies will offer this insurance (some only for those with a valid international
driver’s license) while local companies may or may not. You may wish to request a copy
of their insurance policy and ensure that it states “For Commercial Use”. Regardless, inspect rental vehicles prior to rental and drive with caution, particularly as traffic in Thailand can be quite confusing, especially the habit of Thai motorcycles drivers to drive on
the wrong side of the road. Motorbike Rental: For around 150 to 300 baht per day you
can hire your own 100-150cc motorbike, which will typically require you to leave your
passport as a deposit. Be sure to inspect bikes prior to rental and drive with extreme caution as rental motorbikes are not normally insured and accidents are frequent. Helmets
are required by Thai law. Motorcycles can be rented from rental agencies located on
Rasada Road or from different operators at various beaches.Motorcycle taxis: It costs
approximately 20 baht / person / trip to travel via motorbike taxi around Phuket Town.
Songtaew and Tuk-Tuk: Songtaews are operated along Ranong Road in Phuket Town
to various destinations including most beaches. The cost ranges from 20 baht to 25 baht
/ person / trip. Normally the service is provided from 7.00 a.m. until 5.00 p.m. Tuk-Tuks
can be chartered for travel between the beaches and Phuket Town or between different
beaches; however, rates are negotiable and will cost at least 200 baht to Patong Beach,
230 baht to Karon and Kata Beaches and 300 baht to Nai Han and Kamala Beaches.
Within Phuket Town, Tuk-Tuks should cost 20 baht for short distances. Taxi Meter:
For a Taxi Meter in Phuket, Visitors can call 076 232157-8 to get a metered taxi that will
take them anywhere on Phuket. The metered fare will include a 20 baht surcharge. Boat
242
to islands nearby Phuket: Boats to nearby islands can be found at the following ports:
Rawai Beach: An old local port, it is from here that long-tail boats depart for nearby islands such as Koh He, Koh Racha Yai, Koh Mai Thon, Koh Lon, etc. The chartered price
depends on the distance. Ao Chalong: The largest port of Phuket servicing all kinds of
boats, including cruisers of tour companies that organize package tours to other islands.
Ao Makham: Located near Phanwa Cape, this port is only for cruisers and container
ships. Boat Lagoon Port (Ao Sapam): This port is for traveling boats of tour companies.
Weather
Phuket has a tropical monsoonal climate. It’s warm all year round, but the two periods
of April-May and September-October are the hottest. The September-October period
is also the wettest due to the southwest monsoon. Phuket
is blessed by being in a temperate zone. Even though typhoons and tropical storms occasionally batter Hong Kong
and the Philippines, Phuket enjoys mild weather and while
the heavier rains in September and October can disrupt
things a little, it’s rare that the island has several consecutive days of heavy rain. The best time to visit Phuket is during the cool NE monsoon
season, from December through March, when it isn’t so humid, and the cool breezes keep things comfortable. The average temperature is around 75F to 89F (24C to
32C). Note that lately, as almost everywhere in the world, weather is unpredictable.
Useful Information and Tips
For the Useful Information and Tips, Check out the following for the lowdown
on Thai festivals, money matters, cultural sensitivities, communications, tipping, what
to do on arrival, language, security and emergencies, time-zones, internet access,
electricity and where to go for more info. All you need for a great stay in Phuket!
• http://www.tourismthailand.or.th (Tourism Authority of Thailand)
• http://www.phuket.com (Phuket Tourist)
• http://wikitravel.org/en/Phuket (Wikitravel: Phuket)
243
Transportation to Phuket
By Domestic Airports: Bangkok’s Suvarnabhumi Airport (BKK) serves as the primary international airport in Thailand for numerous international airlines, most with direct flights from abroad landing in the Thai capital. However, some chartered flights and
international service from nearby Asian nations may land at one of the other, smaller international airports within Thailand, such as Phuket (HKT) too. Suvarnabhumi Airport
is Thailand’s premier international air travel gateway and links all aspects of air travel
and ground transport. It also supports the country’s travel and tourism development, as
befits its auspicious name, “Suvarnabhumi”, (Golden Land; pronounced “sue-wannapoom”), which was bestowed upon it by His Majesty King Bhumibol Adulyadej. Airport Rail Link provides train service at the maximum speed of 160 kilometer per hour
on an elevated track parallel to the eastern railway, which covers the distance of approximately 28 kilometers, passing through 8 stations; Phyathai Station, Rajprarop Station,
Makkasan Station and City Air Terminal, Ramkhamhaeng Station, Hua Mark Station,
Thab Chang Station, Lad Krabang Station, and Suvarnabhumi Station.Passengers who
wish to travel to Suvarnabhumi Airport have two options of service: 1. SA Express,
a train service that transports passenger from City Air Terminal or Makkasan Station
to Suvarnabhumi Airport within 15 minutes without stopping at any station along the
way. 2. City Line, a train service that transports passengers between Phyathai Station
and Suvarnabhumi Airport within 30 minutes and stops at every station long the way.
For more information, please enter http://airportraillink.railway.co.th. Flying to Phuket
is arguably the easiest way to get to the island. Most domestic airlines operate several
flights daily between Phuket and Bangkok, Samui, and Chiang Mai. Some domestic
airlines operate flights from Phuket to Surat Thani, Nakhon Si Thammarat, Trang, and
Hat Yai. There are also numerous international airlines that fly directly to Phuket from
various cities around Asia, including Kuala Lumpur, Hong Kong, Penang, Singapore,
and Sydney. Transportation to and from the airport Phuket International Airport is located approximately 30 kilometers north of Phuket Town. Taxis between Phuket Town
244
and the airport cost approximately 400 baht, but the fares to the beaches range between
500 and 600 baht. Minivans charge approximately 80 baht /person to town, but 120
baht/person to Patong, Kata, and Karon Beaches. Phuket Limousine (tel. 076 248596),
located approximately 1 kilometer west of the city, operates hourly shuttles to the airport from 6.30 a.m. to 7.30 p.m. By Train: There is no direct train service to Phuket.
Travelers arriving by train must get off at Phun Phin Railway Station in Surat Thani
Province and continue by regular bus to Phuket. For more information, call the State
Railway of Thailand, 1690, 0 2223 7010, or 0 2223 7020 or visit www.railway.co.th. By
Bus & Coach: When selecting a bus from Bangkok for a long-distance voyage, note
that Thai busses range from the luxurious, towering Super VIP busses to the very colorfully painted express and local busses, which tend to be about the same size as North
American school buses. Each different class of Thailand bus provides different levels of
comfort, some with no onboard toilet and some with full amenities and reclining seats.
Depending on your budget and the length of your trip. To Phuket, Air-conditioned and
non air-conditioned busses leave Bangkok’s Southern Bus Terminal for Phuket several
times daily. Trips by air-conditioned bus, which normally leave in the evening, take
about 13 hours. Call 0 2434 7192, 0 2435 1199 or visit www.transport.co.th for more
information. There are also regular bus services (VIP, air-conditioned, and non-air-conditioned) between Phuket and neighboring provinces such as Krabi, Phang Nga, Chumphon, Koh Samui (bus/boat), Nakhon Si Thammarat, Ranong, Surat Thani, Satun, Hat
Yai, Takua-Pa, and Trang. Departures are from the Phuket Bus Terminal off Phang
Nga Road. For more up-to-date schedules and fares, call Phuket Air-conditioned Bus
Station, tel. 0 7621 1977. By Car: From Bangkok, take Highway No.4 (Petchakasem
Road) through Petchaburi, Prachuap Khiri Khan, Chumphon, Surat Thani and Phang
Nga Provinces, then cross the Thep Krasattri Bridge or Sarasin Bridge to Phuket Island.
The total distance is 862 kilometers and the travel time is approximately 12 hours.
245
About Conference Venue (Hotel)
Hilton Phuket Arcadia Resort & Spa
Hilton Phuket Arcadia Resort & Spa is an idyllic 75-acre resort in Phuket, next to
Karon Beach and 45 minutes from Phuket International Airport. Set in lush, tropical
gardens surrounded by mountains, waterfalls and golden sandy beaches, this Phuket
resort has the largest number of rooms in Southern
Thailand. Wake up in a bright and airy guest room,
decorated in Thai colors, and enjoy a refreshing
drink on the balcony overlooking tropical greenery
or the Andaman Sea. All guest rooms at this beautiful Phuket resort offer WiFi (fees apply) and separate seating areas and desks. Choose a suite for extra
space. A babysitting service, accessible rooms and 24-hour room service are available.
Dine in style at this Phuket hotel, boasting a choice of ten restaurants and bars. Enjoy nightly themed buffets at the Sails restaurant or snacks at Caffe Cino, overlooking the garden pool. Celebrate your wedding at this exotic Phuket resort with excellent event space, including the Grand Ballroom and nine meeting rooms for 5-1,200.
Hold your party in the resort grounds or under a marquee for
up to 1,200 guests. Enjoy exceptional recreational facilities
including 24-hour fitness center, 15 spa treatment rooms, five
outdoor swimming pools, three tennis courts, sauna, steam
room and whirlpool. Ask about a free introductory scuba diving course, play water volleyball, or indulge in a Thai massage in the spa. hotel is now in the TripAdvisor Hall of Fame! staff takes great pride
in offering an exceptional guest experience and being recognized with the TripAdvisor Certificate of Excellence for outstanding traveler reviews five years in a row.
246
Accommodation
Enjoy your stay in a choice of clean and smart rooms at Hilton Phuket Arcadia
Resort & Spa. Our rooms are designed with comfort and convenience in mind; providing a place to unwind or to catch up on work. Explore the various types of rooms, explore the amenities, and choose the space that’s right for you.
Suits: Our suites are a home away from home with a
large sitting room, master bedroom with king size bed,
spacious bathroom with walk-in shower and luxurious tub. Enjoy sweeping views
of the Andaman Sea or the lush tropical gardens from the private balcony. Relax with a drink in the sitting area with friends, watch in-room movies, or connect
and surf the web at the workstation with high-speed internet access (fees apply).
Deluxe Plus Room: Take in sea views or garden views from the private balcony of
this modern, 44 sq.m/473sq.ft room. Relax in the seating area and watch cable TV on
the LCD flat-screen. Catch up on work at the desk with high-speed internet access.
Deluxe Room: Wake up to sea views or garden views and Thai style decor in this
44 sq.m/473sq.ft room. Curl up on the easy chair or get to work at the desk with
high-speed internet access. This spacious and airy room has sliding patio doors.
Location
Hilton Phuket Arcadia Resort & Spa
Address: 333 Patak Road, Karon Beach,
Muang, Phuket, 83100, THAILAND
TEL: +66-76-396433, FAX: +66-76-396136
http://www3.hilton.com, [email protected]
From the airport, take route 4027 to Heroines Monument. Turn right on the bypass and go straight to Chao Fa Road until Chalong Circle. Take route 4028
to Kata/Karon Beach and pass Thavorn Palm Beach. The Hilton Phuket Arcadia Resort & Spa is on the left. Latitude in Decimal is 7.8386743, 98.2950443
247

Program Book - Artificial Intelligence Association of Thailand (AIAT)

Transcription

Similar documents

Annonce S1

Big data Web sites: 40 billion indexed web pages Youtube: 100 hrs

July 10, 2016 - Church of St. Joseph

July-August 2016 - Dallas Fort Worth Metro Golden Retriever Club

LOVE GOD, LOVE OTHERS, SERVE THE WORLD

Press Kit

Judge denies Bevin`s request to close Lexington

corvette z06-rdr

Budget Work Session - Albemarle County Public Schools

to view 2016 Cruise Booklet - Seawanhaka Corinthian Yacht Club