Sparsity methods in systems and control (Sept.19th 14:10-17:30)
Speaker:Prof. Masaaki Nagahara Affiliation:The University of Kitakyushu
Recently, sparsity has been playing a central role in signal processing,
machine learning, and data science,
where a fundamental problem consists in reconstructing (or learning)
a signal (or a function) from observed data,
which may be under-sampled and disturbed by noise.
To address this problem, a method called sparse modeling
has become a hot topic.
In this talk, I will give a brief introduction to sparse modeling
and its applications to systems and control.
In particular, I will give an introduction to "maximum hands-off control,"
which has the minimum support length among all feasible solutions
for saving energy and reducing CO2 emissions
in control systems.
Masaaki Nagahara received the Bachelor's degree in
engineering from Kobe University in 1998,
the Master's degree and the Doctoral degree in informatics from Kyoto
University in 2000 and 2003,
He is currently a full professor at Institute of Environmental Science
the University of Kitakyushu. He is also a visiting professor at
Indian Institute of Technology (IIT) Bombay
since 2017.His research interests include control theory, machine
learning, and sparse modeling.
Dr. Masaaki Nagahara received Young Authors Award in 1999 and Best
Paper Award in 2012 from SICE,
Transition to Practice Award from IEEE Control Systems Society in 2012,
Best Tutorial Paper Award from IEICE Communications Society in 2014,
and Best Book Authors Award from SICE in 2016. He is a senior member of
IEEE, and a member of SICE, ISCIE, IEICE, and JSAI.
Culture Aware Robotic Assistance for the Elderly (Sept.19th 14:10-17:30)
The groundbreaking objective of this workshop is to propose culturally competent care robots, able to autonomously re-configure their way of acting and speaking, when offering a service, to match the culture, customs, and etiquette of the person they are assisting.
By designing robots that are more sensitive to the user’s needs, our innovative solution will offer elderly clients a safe, reliable and intuitive system to foster their independence and autonomy, with a greater impact on quality of life, a reduced caregiver burden, and an improved efficiency and efficacy.
The need for cultural competence in health-care have been widely investigated in the nursing literature.
The study of Transcultural Nursing and Cultural Competence plays a crucial role in providing culturally appropriate nursing care, and it is supported by dedicated International Journals and worldwide associations.
In Europe, Transcultural Nursing is promoted by the European Transcultural Nursing Association (http://europeantransculturalnurses.eu/).
In Japan, to the best of our knowledge, a similar association does not yet exist.
However, as the number of foreign residents in Japan is significantly increasing, the role of cultural competence in health-care will be paid more and more attention by professionals in the fields as well as public decision-makers: then, our solution is expected to contribute to fostering Transcultural Studies in Japan.
In spite of its crucial importance, the commitment of providing assistive systems with cultural competence has been almost totally neglected by researchers.
Today it is technically conceivable to build robots operating within a smart ICT environment that reliably accomplish basic assistive services.
However, state-of-the-art robots consider only the problem of “what to do” in order to provide a service: they produce rigid recipes, which are invariant with respect to the place, person and culture.
This workshop stems from the consideration that reasons about “what to do” is not sufficient and necessarily doomed to fail ? just like cultural competence is crucial for human caregivers:
? from the user’s perspective, a culturally appropriate behavior is key to improve acceptability;
? from the commercial perspective, it will open new avenues for marketing robots across different countries.
To achieve its groundbreaking objective, this workshop will follow a three-step approach. First, we will study how to represent cultural models, how to use these models in sensing, planning and acting, and how to acquire them.
Second, we will consider three (physically identical) replicas of a commercial robot on the market and integrate cultural models into them, by making them culturally competent.
Third, we will test the three robots, customized for three different cultures, in two test sites in EU and Japan, on a number of elderly volunteers and their informal caregivers.
This workshop will adopt a robust user-testing approach, by performing the same level of testing in EU (two different cultural groups) and in Japan (one cultural group).
Rapid demographic change constitutes an unprecedented societal challenge for Japan.
I will introduce a new EC Horizon 2020 project “CARESSES”, aiming at developing culturally competent elderly care robots, jointly commissioned by the Ministry of Internal Affairs and Communications of Japan.
We envision a future in an aging society, where robots are able to autonomously re-configure their way of acting and speaking, when offering a service, to match the culture, customs, and etiquette of the person they are assisting.
By designing robots that are more sensitive to the user’s needs, our innovative solution will offer elderly clients a safe, reliable, and intuitive system to foster their independence and autonomy, with a greater impact on quality of life, a reduced caregiver burden, and an improved efficiency and efficacy.
The preliminary results of the initial phase of project and its final goals will be presented.
Nak Young Chong received the B.S., M.S., and Ph.D. degrees in mechanical engineering from Hanyang University, Seoul, Korea, in 1987, 1989, and 1994, respectively.
From 1994 to 2007, he was a member of research staff at Daewoo Heavy Industries and KIST in Korea, and MEL and AIST in Japan.
In 2003, he joined the faculty of Japan Advanced Institute of Science and Technology, where he currently is a Professor of Information Science and directs the Robotics Laboratory.
He was a Visiting Professor at Northwestern University, Georgia Institute of Technology, University of Genoa, and Carnegie Mellon University, and also served as an Associate Graduate Faculty at UNLV.
He serves/served as Editor of the IEEE Robotics and Automation Letters, Sage International Journal of Advanced Robotic Systems, IEEE ICRA CEB, and IEEE CASE CEB, and Associate Editor of the IEEE Transactions on Robotics and Springer Journal of Intelligent Service Robotics.
He served as Program Chair/Co-Chair for JCK Robotics 2009, ICAM 2010, IEEE Ro-Man 2011, IEEE CASE 2012, IEEE Ro-Man 2013, URAI 2012/2013, and DARS 2014.
He also served as Co-Chair for IEEE-RAS Networked Robots Technical Committee from 2004 to 2006, and Fujitsu Scientific System Working Group from 2004 to 2008. He is the General Co-Chair of URAI 2017.
Previous research revealed that social activities, physical health, and personality are respectively strong predictors on well-being of the elderly.
In this talk, we report the comprehensive relationships of these factors on well-being such as satisfaction of life, autonomy and positive
relationships of the elderly in Japan. We discuss the possibility of the robotics to support and
enhance well-being of the elderly.
Hiroko Kamide received the M.S degree in Human Sciences from Osaka Univ., Japan, in 2005, and PhD. Degree in Human Science from Osaka Univ., Japan, in 2008. Since April 2007, she has been a Research Fellow of the Japan Society for the Promotion (DC2) and since April 2008, she has been a Research Fellow of the Japan Society for the Promotion (PD).
Since December 2009, she has been a Specially Appointed Assistant Professor with the Department of Systems Innovation, Graduate School of Engineering Science, Osaka University, Osaka, Japan.
In 2015 she moved to Tohoku University in Japan as an Assistant Professor and since 2016 she has been a designated Associate Professor in Nagoya University in Japan.
Her research interests include well-being, psychological evaluation of robots, and human-robot interaction. Dr. Kamide is a member of the IEEE Robotics and Automation Society, Robotics Society of Japan, The Japanese Psychological Association, The Japanese Society of Social Psychology, and Society for Personality and Social Psychology.
Ambient Assisted Living (AAL) pursues the issues of how information and communication technology (ICT) is used to support the quality of life of the elderly and disabled people in smart home environment.
Among the many AAL solutions, universAAL is one of the most open platforms that provides a holistic and standardized approach for an easy and economic development of AAL services.
However, the compatibility with smart home networking systems that are based on ECHONET standard is not supported.
Thus, this research elucidates the bridging in between the universAAL and the ECHONET standard and thereby enable the attainment of novel AAL applications and services for ECHONET-based smart home automation systems.
Yuto Lim received the B.Eng. (Hons) and M.Inf. Technology degrees from Universiti Malaysia Sarawak (UNIMAS), Malaysia in 1998 and 2000, respectively.
He received the Ph.D. degree in communications and computer engineering from Kyoto University in 2005.
He was a visiting researcher at Fudan University in China for two months. During 2005-2009, he was an expert researcher at National Institute of Information and Communications Technology (NICT), Japan.
Since 2009, he has been working at Japan Advanced Institute of Science and Technology (JAIST) as an associate professor.
His research interests include multihop wireless networks, wireless sensor networks, home networks, wireless mesh networks, heterogeneous wireless networks, network coding, cyber-physical system.
He is a member of IEEE, IEICE and IPSJ.
Generating emotional body expressions for socially assistive robots has been gaining increased attention to enhance the engagement and empathy in human-robot interaction.
In this presentation, I introduced a new model of emotional body expression for the robot inspired by social and emotional development of infant from their parents.
An infant is often influenced by social referencing, meaning that they perceive their parents' interpretation about emotional situations to form their own interpretation.
Similar to the infant development case, robots can be designed to generate representative emotional behaviors by clustering various emotional behavior samples from human partners.
To realize this idea, we used self-organized neural networks to extract the human’s representative emotional behaviors and the extracted behaviors are transformed by mapping functions to generate an appropriate robot’s real emotional behaviors.
A public human action dataset and the Pepper humanoid robot were used to validate the proposed emotional behavior expression method.
Sungmoon Jeong received the B.S., M.S., and Ph.D. degrees from Kyungpook National University, Korea in 2006, 2008 and 2013, and then he joined Japan Advanced Institute of Science and Technology (JAIST) as an assistant professor.
He has received the “Young Researcher award” from Asia Pacific Neural Network Assembly (APNNA), 2013 and “Young Scientist award” from Korea Robotics Society (KROS), 2014.
He is an Associate Editor of IEEE International Conference Robotics and Automation (ICRA) from 2014 and program committee members of International Conference on Neural Information Processing (ICONIP) and Human Agent Interaction (HAI) from 2013.
His scientific interests consist of developmental learning, computational vision research, machine learning, pattern recognition, neuro-robotics and their application systems such as autonomous vehicles.
Social signal processing (SSP) is a research area that aims at developing the computational model to recognize social behavior such as turn taking, personality, communicative competence, and communication skills from multimodal information including vocal activity, prosody, gazes, gesture and posture.
SSP contributes many kinds of application such as human robot interaction and autonomous conversation analysis.
In this talk, the fundamental techniques of SSP and ongoing studies using SSP are introduced.
Shogo Okada is an associate professor at the Japan Advanced Institute of Science and Technology (JAIST) in Japan.
He obtained his Ph.D. in 2008 from Tokyo Institute of Technology in Japan.
In 2008, 2011, he joined the Kyoto University, Tokyo Institute of Technology as an assistant professor.
He visited IDIAP research institute, Switzerland as a visiting faculty in 2014. His research interests include social signal processing, data mining for human communication.
He is a member of the ACM, IEICE, JSAI.
Robots are expected to act as mediators to elicit more active communication and provide life support for humans.
Social robots have found a number of applications in many aspects of our daily life, including elderly care and therapeutic purposes (e.g. therapy for children with autism). The critical role of robots here is to interact with and assist humans in their every-day activities.
Considering a wide variety of users, the robots should be also capable of deciding what kind of services and interactions they perform.
The accurate and autonomous evaluation is needed through the technology, especially if the users are children or people with special needs.
For this “user-centered” human-robot interaction, this requires that the social robots can learn the user’s emotional states and be able to respond to it accordingly.
The main topic of this talk is social robots for autism therapy and elderly people.
Jaeryoung Lee received her B.S. in Mechanical Engineering from Pusan National University, Korea in 2009 and her M.S. and Ph.D. in Mechanical Engineering from the Nagoya University, Japan in 2012, 2014.
She is currently an assistant professor of the department of Robotic Science and Technology in Chubu University, Japan, where she teaches Motor Engineering, Humanoid Robot Control, and Human Robot Interaction.
Her research interests are in robot-assisted therapy, rehabilitation robotics, and human machine interaction She is currently the PI of MEXT project regarding robots for autism, and a partner in EU & MIC project Culture Aware Robots and Environmental Sensor Systems for Elderly Support, and so on.
She received the Research Award at Society of Life Support Engineering 2014 and the Best Paper Award at IEEE MHS 2015.
She organized several highly successful workshops and organized sessions on social robots.
Artificial Intelligence for Games (Sept.19th 14:10-17:30)
Organizer(s): Kokolo Ikeda (Japan Advanced Institute of Science and Technology)
Artificial Intelligence (AI) methods such as machine learning or optimization are now widely used for human society and still attracting many researchers, including SICE community.
Games are often employed as the testbed of AI methods, since games are well-defined, easy to understand, simple but difficult to play well.
Many AI methods have been proposed, sophisticated through games, and applied to other problems.
This workshop will presents the recent progress and future directions of AIs for games, by top researchers.
The target game is not limited to classical board-games such as Chess or Go, but also Video-games or a party game "Are you a Werewolf?" will be dealt.
The purpose of research is not limited to AI strength, but also human-likeness and entertainment aspect will be discussed. Intended Audience: people who are interested in artificial intelligence methods, computer game players, or games themselves.
1.The history of game AI and recent movement of domestic game AI competitions
Takeshi Ito, The University of Electro-Communications
2.A decade of advances in computer shogi and the Bonanza program
Kunihito Hoki, The University of Electro-Communications
3.The current state of the computer Go technology
Hideki Kato, Team DeepZen and DeepZenGo project
4.Building Computer Mahjong Player
Naoki Mizukami, The University of Tokyo
5.AI Wolf - Artificial Intelligence Based Werewolf Game Player -
Takashi Otsuki, Yamagata University Talk/Lecture 1: The history of game AI and recent movement of domestic game AI competitions
"Make AI play games" This purpose has generated many kinds of research on AI.
Chess had been the center of game research until Deep Blue beat Kasparov in 1997.
Even in Go, which is considered to be the most difficult to develop among famous board games, it is going to exceed the human top player with the advent of AlphaGo.
In this lecture, we first look back on the history of researches such as Checkers, Othello, Chess, Shogi and Go.
And as new game AI research areas attracting attention in recent years, I will introduce the competition movements such as Werewolf, Mini 4WD, TUBSTAP, Curling, etc..
Speaker: Takeshi Ito (The University of Electro-Communications)
Short Biography: Takeshi Ito recieved M.E., and D.E. degrees in Information Engineering from Nagoya University in 1991, 1994.
Then, he became a research associate of the Department of Computer Science, the University of Electro-Communications in 1994, an assistant professor in 2007.
He is a director of the Cognitive Science and Entertainment Research Station in the University of Electro-Communications from 2006.
His research interests is cognitive science approach on human thought processes in playing games.
Talk/Lecture 2: A decade of advances in computer shogi and the Bonanza program
Shogi is one of the major variants of chess. The material balance is less important and the branching factor is greater than in Western chess.
Because of these properties, creating a decent computer shogi player was a difficult challenge.
When Deep Blue defeated the World Chess Champion Garry Kasparov in 1997, the most advanced Shogi program at that time was no better than an ordinary amateur player.
In this decade, computer shogi programs have started to defeat human experts.
The program Akara2010 defeated one of the top female shogi professionals in 2010, another program, Ponanza, achieved the highest rating in blitz games among all players in the most popular online shogi server, Shogi Club 24, in 2011,
and Bonkras defeated a retired top professional player, Yonenaga, in 2012. In this talk, two main features of the computer shogi program Bonanza, one of the famous shogi programs, will be presented.
One main feature is the optimization method used to train the full evaluation function with more than ten million parameters.
The other is a game tree search method which is more brute-force and less selective than other popular approaches.
Speaker: Kunihito Hoki (The University of Electro-Communications)
Short Biography: Kunihito Hoki (The University of Electro-Communications) Kunihito Hoki received his Ph.D. from the Graduate School of Science of Tohoku University in Miyagi, Japan.
He is presently working in the department of computer and network engineering, graduate school of informatics and engineering at the University of Electro-Communications in Tokyo, Japan.
His current research interests include computer games and machine learning.
Talk/Lecture 3: The current state of the computer Go technology
Deep learning has boosted image recognition performance a lot.
Deep learning can also be applied optimal planning problems, including computer games.
Google/DeepMind's AlphaGo beat Ke Jie, the world strongest Go player with 3-0 in May 2017.
I'll talk about how and why deep learning has made computer Go programs stronger than top level professionals and what weakpoints are left.
Speaker: Hideki Kato (Team DeepZen and DeepZenGo project)
Short Biography: Hideki Kato was born in 1953. He received MS from Tokyo Institute of Technology.
He was working on applied AI at Fujitsu Laboratories Ltd. from 1982 to 2001.
He started the development of computer Go programs at 2006.
He is the representatives of Team DeepZen and DeepZenGo project since 2009 and 2016, resp.
He is the members of IEICE, IPSJ, JNNS, ICGA.
Talk/Lecture 4: Naoki Mizukami (The University of Tokyo)
Mahjong is a traditional and the most popular table games in Japan.
From an AI research point of view, Mahjong is a challenging game because (a) it is played with more than two players, (b) it is an imperfect information game, and (c) the number of information sets is much bigger than those of popular card games such as poker.
In this lecture, we describe how to build the state of the art computer mahjong player using supervised learning, Monte Carlo simulation and reinforcement learning.
Speaker: Naoki Mizukami (The University of Tokyo)
Short Biography: Naoki Mizukami recieved B.E., M.E. degrees in Information Engineering from Kanazawa University, The University of Tokyo in 2013, 2015.
Talk/Lecture 5: AI Wolf - Artificial Intelligence Based Werewolf Game Player -
Artificial intelligence (AI) has been tested by the board games such as chess, shogi and Go, that of the kind called "complete information game" in which players can see what their opponents are doing. Now that AlphaGo, the world's most famous AI Go program, has defeated Chinese grandmaster, it is said that the next step of game-playing AI is "incomplete information games" such as poker, bridge and mahjongg, that have information hidden from players.
From the communicative point of view, while the conversations between players are important in such games, especially in "communication games" that are fully conducted through conversations, relatively few studies focus on the application of AI in communication games.
Therefore, we present "Are You a Werewolf?" which is one of the most popular communication games with incomplete information, as a new standard game problem in the AI field, and started a project to create an "AIWolf" which is an AI agent which can play Werewolf game in place of a human player.
Our project consists of multiple research teams because AIWolf requires many research areas such as design of game playing AI, analysis of the human player, natural language processing, agent technology and human-agent interaction, etc.
In order to solve the problems in these areas, we adopt a collective intelligence approach which uses the competition like RoboCup Soccer, providing a common platform to develop an AIWolf agent for a competition which realizes the collective intelligence with participants.
We have held the 1st and 2nd Werewolf Intelligence Competition (WIC) at the Computer Entertainment Developers Conference (CEDEC) on August in 2015 and 2016 respectively.
About forty agents, that could play the games without errors, participated in the preliminary competition, and top-fifteen in winning percentage (WPCT) survived to the final.
In the final, a large number of games, for example, 1,124,890 games in the 1st WIC, were played, and the agents were ranked in order of their WPCT.
Our findings obtained through the two WICs suggest that the competition facilitates a collective intelligence approach to achievement of AIWolf we aim to develop.
The source codes of the agents are open-source software and available on our site (http://aiwolf.otg/).
for developer's utilizing to make stronger agent which can beat the former champion.
In the past competitions, to ease the development, the agents communicate with others using "AIWolf Protocol" which consists of limited kinds of shortened utterance.
In the 3rd WIC this year, we organize the first competition of agents which communicate in natural language, for stepping forward to the agent which can play Werewolf game with humans.
Speaker: Takashi Otsuki, Yamagata University
Short Biography: Takashi Otsuki received the B.E., M.E. and Dr.Eng. degrees in information engineering from Tohoku University, Sendai in 1988, 1990 and 1993, respectively.
He is currently a associate professor of Academic Assembly, Yamagata University.
He has been engaged in research on statistical language model, speech information processing and intelligent robotics.
His research interests are pattern recognition and artificial intelligence.
In AIWolf project, he is in charge of the development of the platform. He also develop .NET version AIWolf platform voluntarily.
He is a member of the Information Processing Society of Japan, Japanese Society for Artificial Intelligence and the Acoustical Society of Japan.