|
[ Return to top ]
Klaus-Robert Müller Toward Brain Computer Interfacing Abstract Brain Computer Interfacing (BCI) aims at making use of brain signals for e.g. the control of objects, spelling, gaming and so on. This talk will first provide a brief overview of Brain Computer Interface from a machine learning and signal processing perspective. In particular it shows the wealth, the complexity and the difficulties of the data available, a truely enormous challenge: In real-time a multi-variate very strongly noise contaminated data stream is to be processed and neuroelectric activities are to be accurately differentiated. Finally, I report in more detail about the Berlin Brain Computer (BBCI) Interface that is based on EEG signals and take the audience all the way from the measured signal, the preprocessing and filtering, the classification to the respective application. BCI as a new channel for man-machine communication is discussed in a clinical setting and for gaming. This is joint work with Benjamin Blankertz, Michael Tangermann, Matthias Krauledat, Claudia Sanelli, Stefan Hauffe (TU, Berlin) and Gabriel Curio (Charite, Berlin). Biography Klaus-Robert
Müller received the Diplom degree in mathematical physics 1989 and the
Ph.D. in theoretical computer science in 1992, both from University of
Karlsruhe, Germany. From 1992 to 1994 he worked as a Postdoctoral fellow
at GMD FIRST, in Berlin where he started to built up the intelligent
data analysis (IDA)group. From 1994 to 1995 he was a European Community
STP Research Fellow at University of Tokyo in Prof. Amari's Lab. From
1995 on he is department head of the IDA group at GMD FIRST (since 2001
Fraunhofer FIRST) in Berlin and since 1999 he holds a joint associate
Professor position of GMD and University of Potsdam. In 2003 he became
full professor at University of Potsdam, in 2006 he became chair of the
machine learning department at TU Berlin. He has been lecturing at
Humboldt University, Technical University Berlin and University of
Potsdam. In 1999 he received the annual national prize for pattern
recognition (Olympus Prize) awarded by the German pattern recognition
society DAGM and in 2006 the SEL Alcatel communication award. He serves
in the editorial board of Computational Statistics, IEEE Transactions on
Biomedical Engineering, Journal of Machine Learning Research and in
program and organization committees of various international
conferences. His research areas include statistical learning theory for
neural networks, support vector machines and ensemble learning
techniques. He contributed to the field of signal processing working on
time-series analysis, statistical denoising methods and blind source
separation. His present application interests are expanded to the
analysis of biomedical data, most recently to brain computer interfacing
and genomic data analysis. [ Return to top ]
Lei Xu
Abstract There are two key challenges for statistical learning. One is finding appropriate mathematical representations to suit various dependence structures underlying the world, for which many learning models have been studied in past decades. The other is getting a good theory for seeking a model with an appropriate sale or complexity to learn reliable structures underlying a finite size of samples. Conventionally, a number of candidate models in different scales are enumerated, with unknown parameters estimated under the maximum likelihood (ML) principle. Thereafter, one of typical learning theories, being different from ML, is applied to select the candidate in a best scale. However, not only this two-phase implementation needs a vast computing cost, but also each of these typical approaches can provide a rough estimate only. Bayesian Ying Yang (BYY) system jointly considers two types of learning for interpreting what are observed from its world and for skills of solving problems encountered in the world, which provides a general framework for a number of existing typical learning models. The best harmony principle provides a general guideline for making parameter learning and model selection jointly. Particularly, the best Ying-Yang harmony leads to not only a criterion that outperforms typical model selection criteria in a two-phase implementation, but also an automatic model selection on several typical learning tasks with an appropriate model scale obtained automatically during parameter learning while with computing cost saved significantly. Also, degenerated cases return to several existing theories, e.g., AIC and variants, marginal likelihood type Bayesian (BIC, MDL, etc), variational Bayes. This talk consists of two parts. The first provides an introduction of BYY system and best harmony learning, with links to several existing learning models and theories. The second part introduces further details on BYY systems with its components featured by Gaussian manifolds, including Gaussian mixture, local factor analysis (LFA), temporal LFA and its HMM gated extensions, etc, with experimental results on several typical problems in machine learning and pattern recognition. Biography Lei Xu is a chair professor of Chinese Univ Hong Kong (2002-), a Chang Jiang Chair Professor of Peking Univ, an IEEE Fellow (2001-) and a Fellow of International Association for Pattern Recognition (2002-), and a member of European Academy of Sciences (2002-). He completed his PhD thesis at Tsinghua Univ by the end of 1986, then joined Peking Univ in 1987, and further promoted exceptionally to an associate professor in 1988. During 1989-93, he worked at several universities in Europe and North America, including Harvard and MIT. He joined CUHK in 1993 as senior lecturer, became professor in 1996 and took the current position since 2002. Prof. Xu has published a number of well-cited papers in the literatures of neural networks, statistical learning, and pattern recognition, e.g., his papers got over 1800 citations according to SCI-Expended (SCI-E) and over 3600 citations according to Google Scholar (GS), with his 10 most frequently cited papers scored near 1100 (SCI-E) and 2500 (GS). Among them, one single his paper has scored 360 (347+13) (SCI-E) and 932 (776+113+43) (GS). He served as associate editor for several journals and as general chair or program committee chair of a number of international conferences. He also served as a past governor of international neural network society (INNS), a past president of Asian-Pacific Neural Networks Assembly (APNNA), and a member of Fellow committee of IEEE Computational Intelligence Society, as well as a nominator for Kyoto prize. Moreover, he has received several Chinese national academic awards (including 1993 National Nature Science Award) and international awards (including 1995 INNS Leadership Award and the 2006 APNNA Outstanding Achievement Award).
[ Return to top ]
Shiro Usui
Abstract To promote
international cooperation in the new field of Neuroinformatics (NI), the
Neuroinformatics Japan Center (NIJC) at RIKEN Brain Science Institute
(BSI) in Wako, Japan has been established in 2005 as the Japan Node
(J-Node) for coordination with the International Neuroinformatics
Coordinating Facility (INCF). The Laboratory for Neuroinformatics was
established in 2002 at RIKEN BSI, and created an NI base-platform
"XooNIps" following the concepts and experience from constructing the Visiome Platform(VP), which is developed under the Neuroinformatics
Research in Vision (NRV) Project. XooNIps features better scalability,
extensibility, and customizability to operate under various site
policies in the general NI community and can be easily customized to
support different databases and portals. It provides a framework for
successfully accumulating, sharing and making public resources which
were once difficult to accumulate, share and make public. Based on VP,
nine J-Node Platforms have been developed by NIJC platform committees
from selected research areas utilizing XooNIps. XooNIps contributes not
only in NI field but in such diverse areas as library depositories and
university research resources. Biography Shiro Usui graduated from the University of California at Berkeley in 1974 and obtained his PhD in electrical engineering and computer science. He then became a research assistant at Nagoya University. He moved to Toyohashi University of Technology in 1979 as a lecturer, and has been a professor since 1986. In 2003 he moved to the RIKEN Brain Science Institute as the head of Neuroinformatics Laboratory, and became the Director of the Neuroinformatics Japan Center in 2007. His research interests are Neuroinformatics, computational neuroscience and physiological engineering in vision science . He is the author of Neuroinformatics, Mathematical Models of Brain and Neural Systems, and several other books. He is a fellow of the IEEE and the IEICE and was the president of the Japanese Neural Network Society for the years of 2005 and 2006.
[ Return to top ]
DeLiang Wang Cocktail Party Processing Abstract Speech segregation, or the cocktail party problem, has proven to be extremely challenging. This presentation describes a computational auditory scene analysis (CASA) approach to the cocktail party problem. This monaural approach performs auditory segmentation and grouping in a two-dimensional time-frequency representation that encodes proximity in frequency and time, periodicity, amplitude modulation, and onset/offset. In segmentation, our model decomposes the input mixture into contiguous time-frequency segments. Grouping is first performed for voiced speech where detected pitch contours are used to group voiced segments into a target stream and the background. In grouping voiced speech, resolved and unresolved harmonics are dealt with differently. Grouping of unvoiced segments is based on the Bayesian classification of acoustic-phonetic features. This CASA approach has led to major advances towards solving the cocktail party problem. Biography DeLiang Wang
received the B.S. degree in 1983 and the M.S. degree in 1986 from Peking
(Beijing) University and the Ph.D. degree in 1991 from the University of
Southern California. Since 1991, he has been with the Department of
Computer Science & Engineering and the Center for Cognitive Science at
The Ohio State University, where he is a Professor. He has also been a
visiting scholar at Harvard University and Oticon A/S. He received the
U.S. National Science Foundation Research Initiation Award in 1992 and
the U.S. Office of Naval Research Young Investigator Award in 1996. He
is the recipient of the 2008 Helmholtz Award from the International
Neural Network Society for his contributions in machine perception. His
paper: "The time dimension for scene analysis" received the 2005
Outstanding Paper Award of IEEE Transactions on Neural Networks. He is
an IEEE Fellow.
[ Return to top ]
Kristin P. Bennett Optimization and Machine Learning Abstract In this talk we examine the interplay of optimization and machine learning. Great progress has been made in machine learning by cleverly reducing machine learning problems to convex optimization problems with one or more hyper-parameters. The availability of powerful convex-programming theory and algorithms has enabled a flood of new research in machine learning models and methods. But many of the steps necessary for successful machine learning models fall outside of the convex machine learning paradigm. Thus we now propose framing machine learning problems as Stackelberg games. The resulting bilevel optimization problem allows for efficient systematic search of large numbers of hyper-parameters. We discuss recent progress in solving these bilevel problems and the many interesting optimization challenges that remain. Finally, we investigate the intriguing possibility of novel machine learning models enabled by bilevel programming. Biography Kristin P.
Bennett is a Professor in the Mathematical Sciences and Computer
Sciences Departments at Rensselaer Polytechnic Institute. She is an
active member of the machine learning, data mining, and operations
research communities, serving as present or past associate editor for
ACM Transactions on Knowledge Discovery from Data, SIAM Journal on
Optimization, Naval Research Logistics, Machine Learning Journal, IEEE
Transactions on Neural Networks, and Journal on Machine Learning
Research. She served as program chair of the Eleventh ACM SIGKDD
International Conference on Knowledge Discovery and Data Mining. She has
a Ph.D. and M.S. in Computer Sciences from the University of
Wisconsin-Madison, and a B.S. in Mathematics and Computer Science from
the University of Puget Sound. She has been researching
mathematical-programming approaches to machine learning such as support
vector machines since 1989 with over sixty papers on this subject. In
addition, she has worked extensively on successful application of
machine learning to problems in chemistry, biology, engineering, and
business.
[ Return to top ]
Witold Pedrycz Collaborative Architectures of Fuzzy Modeling Abstract There are rapidly emerging needs to deal with distributed sources of data (sensors and sensor networks, web sites, databases). While recognizing their limited accessibility at a global level (associated with technical constraints and/or privacy issues) and fully acknowledging benefits of collaborative processing, we propose a concept of Collaborative Computational Intelligence (CI), and collaborative fuzzy models, in particular. The variety of possible mechanisms of interaction is organized into a setting of the C3 interaction paradigm (communication - collaboration - consensus). This helps us offer a coherent taxonomy of various schemes of interaction which in the sequel implies a certain development of a suite of algorithms. In this setting, the role granular information in the establishing of the mechanisms of interaction plays a pivotal role. We consider distributed fuzzy models and fuzzy modeling. In particular, we elaborate on the key design issues concerning fuzzy rule-based systems with local functional models occurring at their conclusion parts and show how the fundamental modes of interaction are exploited here. It will be demonstrated that more advanced constructs such as type-2 fuzzy sets emerge naturally in distributed fuzzy modeling and come with a well-defined semantics of their membership functions by being fully reflective of the character of the underlying distributed data. In the context of collaborative fuzzy modeling, we bring forward a concept experience-consistent fuzzy system identification showing how fuzzy models built on a basis of limited data can benefit from taking advantage of the past experience conveyed in the form of previously constructed fuzzy models. Detailed algorithmic considerations embrace several design scenarios in which we apply the mechanism of experience consistency at the level of conditions and conclusions of the rules. We also show that a level of achieved experience-driven consistency can be quantified through fuzzy sets (fuzzy numbers) of the parameters of the local models standing in the conclusion parts of the rules this leading to the emergence of granular constructs of fuzzy modeling. Biography Witold
Pedrycz received the M.Sc., and Ph.D., D.Sci. all from the Silesian
University of Technology, Gliwice, Poland. He is a Professor and Canada
Research Chair (CRC) in Computational Intelligence in the Department of
Electrical and Computer Engineering, University of Alberta, Edmonton,
Canada. Dr. Pedrycz is an IEEE Fellow and IFSA Fellow. His main research
interests encompass fundamentals of Computational Intelligence, Granular
Computing, fuzzy modeling, knowledge discovery and data mining, fuzzy
control including fuzzy controllers, pattern recognition,
knowledge-based neural networks, relational computing, and Software
Engineering. He has published vigorously in these areas. He is an author
of 11 research monographs and over 250 journal papers published in
highly reputable journals. His research is highly cited and he is also
on the list Highly cited researcher on ISI HighlyCited.comSM. Witold
Pedrycz has been a member of numerous program committees of IEEE
conferences in the area of Computational Intelligence, Granular
Computing, fuzzy sets and neurocomputing. He was a Program Chair of the
2007 Int. Conf on Machine Learning and Cybernetics, August 19-22, 2007,
Hong Kong. He was also a General Chair of NAFIPS 2004, June 24-26, 2004,
Banff, Alberta- a flagship conference of the NAFIPS Society. He
currently serves as an Associate Editor of IEEE Transactions on Systems
Man and Cybernetics, IEEE Transactions on Neural Networks, and IEEE
Transactions on Fuzzy Systems. He is also on editorial boards of over 10
international journals. Dr Pedrycz is also an Editor-in-Chief of
Information Sciences and IEEE Transactions on Systems, Man, and
Cybernetics part A (with the term starting in January 2008). Dr. Pedrycz
is the past president of IFSA and the past president of NAFIPS. Dr.
Pedrycz is a recipient of the prestigious Norbert Wiener Award which is
one of the two highest awards of the IEEE Systems, Man, and Cybernetics
Society. He is also a recipient of the K.S. Fu of NAFIPS. [ Return to top ]
Hani Hagras
Abstract Type-1 Fuzzy Logic Controllers (FLCs) have been applied to date with great success to many different applications. However, for many real-world applications, there is a need to cope with large amounts of uncertainties. The traditional type-1 FLC using crisp type-1 fuzzy sets cannot directly handle such uncertainties. A type-2 FLC using type-2 fuzzy sets can handle such uncertainties to produce a better performance. Hence, type-2 FLCs will have the potential to overcome the limitations of type-1 FLCs and produce a new generation of fuzzy controllers with improved performance for many applications, which require handling high levels of uncertainty. Through the review of the various type-2 FLC applications, it has been shown that as the level of imprecision and uncertainty increases, the type-2 FLC will provide a powerful paradigm to handle the high level of uncertainties present in real-world environments. It has been also shown in various applications that the type-2 FLCs have given very good and smooth responses that have always outperformed their type-1 counterparts. Thus, using a type-2 FLC in real-world applications can be a better choice since the amount of uncertainty in real systems most of the time is difficult to estimate. It is envisaged to see a wide spread of type-2 FLCs in many real-world application in the next decade. This talk will introduce the interval type-2 FLCs and how they present a way forward for fuzzy systems in real world environments and applications that face high levels of uncertainties. The talk will present different ways to design interval type-2 FLCs. The talk will also present the successful application of type-2 FLCs to many real world settings including industrial environments, mobile robots, ambient intelligent environments and intelligent decision support systems. The talk will conclude with an overview on the future directions of type-2 FLCs. Biography Hani Hagras is a Professor of Computer Engineering in the German University in Cairo, Egypt. He is also a Professor in the Department of Computing and Electronic Systems, Director of the Computational Intelligence Centre and the Head of the Fuzzy Systems Research Group in the University of Essex, UK. He received the B.Sc. and M.Sc. degrees from the Electrical Engineering Department at Alexandria University, Egypt, and the Ph.D. degree in computer science from the University of Essex, U.K. His major research interests are in computational intelligence, notably type-2 fuzzy systems, fuzzy logic, neural networks, genetic algorithms, and evolutionary computation. His research interests also include ambient intelligence, pervasive computing and intelligent buildings. He is also interested in embedded agents, robotics and intelligent control. He has authored more than 120 papers in international journals, conferences and books. His work has received funding that totalled to about £2 Million in the last five years from the European Union, the UK Department of Trade and Industry (DTI), the UK Engineering and Physical Sciences Research Council (EPSRC), the UK Economic and Social Sciences Research Council (ESRC), the Korea- UK S&T fund as well as several industrial companies. He is a Fellow of the Institution of Engineering and Technology (IET (IEE)) and a Senior Member of the Institute of Electrical and Electronics Engineers (IEEE). He is the Chair of the IEEE CIS Task Force on Intelligent Agents and Co-Chair of the IEEE CIS Task Force on Extensions to Type-1 Fuzzy Sets. His research has won numerous prestigious international awards where most recently he was awarded by the IEEE Computational Intelligence Society (CIS), the Outstanding Paper Award in the IEEE Transactions on Fuzzy Systems. In addition, he was awarded the IET Knowledge Networks Award. He is a member of the IEEE Computational Intelligence Society (CIS) Fuzzy Systems Technical Committee. He is also a member of the IEEE Industrial Electronics Society (IES) Technical Committee of the Building Automation, Control and Management. In addition he is member of the Executive Committee of the IET Robotics and Mechatronics Technical and Professional Network. He is also a member of the International Medical Informatics Association (IMIA) working group on Smart Homes and Ambient Assisted Living. Prof. Hagras chaired several international conferences where most recently he served as the General Co-Chair of the 2007 IEEE International Conference on Fuzzy systems London, July 2007 and he also serves as Programme Chair for the 2008 IET International Conference on Intelligent Environments, Seattle, USA. He served as a member of the international programme committees of numerous international conferences.
[ Return to top ]
Ronald R. Yager
Abstract Two important classes of human behavioral modeling can be readily identified. In the first type of modeling we are trying to digitally model a "human" or "human like" agent that interacts with some more complex environment, which can be digital or real. It is central to the construction of synthetic agents, computational based training systems and machine learning. It is implicit in our attempts to construct intelligent systems. It can be denoted as I-P modeling as an acronym for Individual Participant modeling. A second type of human modeling involves a system of interacting human participants. It is the modeling of social networks. This is often what is done in social sciences. The modeling here is from the perspective of an external observer. Interest in this second type of modeling from the perspective of computational intelligence is of much more recent vintage. This type of modeling has arisen in importance with the wide spread use of the Internet and its role in the fostering of cooperation and social networking. Network modeling is also playing a central role in helping to understand the structure of various criminal and terror organizations. In both types of modeling we require an ability to formally represent sophisticated cognitive concepts that are often at best described in imprecise linguistic terms. Our goal in this talk is to discuss the role that soft computing methods can play in the future development these types of human behavioral modeling. With the aid of a fuzzy set we can formally represent sophisticated imprecise linguistic concepts in a manner that allows for the types of computational manipulation needed for reasoning in behavioral models based on human cognition and conceptualization. With the use of the Dempster-Shafer theory we can provide machinery for including randomness in the fuzzy systems modeling process. This combined methodology provides a framework with which we can construct models that can include both the complex cognitive concepts and unpredictability needed to model human behavior. Furthermore in discussing the qualities of importance in social networks such as political ties, kinship obligations and friendship we use attributes such as intensity, durability and reciprocity. These attributes are most naturally evaluated in imprecise terms. Biography Ronald R. Yager has worked in the area of fuzzy sets and related disciplines of computational intelligence for over twenty-five years. He has published over 500 papers and fifteen books. He was the recipient of the IEEE Computational Intelligence Society Pioneer award in Fuzzy Systems. Dr. Yager is a fellow of the IEEE, the New York Academy of Sciences and the Fuzzy Systems Association. He was given an award by the Polish Academy of Sciences for his contributions. He served at the National Science Foundation as program director in the Information Sciences program. He was a NASA/Stanford visiting fellow and a research associate at the University of California, Berkeley. He has been a lecturer at NATO Advanced Study Institutes. He received his undergraduate degree from the City College of New York and his Ph. D. from the Polytechnic University of New York. Currently, he is Director of the Machine Intelligence Institute and Professor of Information and Decision Technologies at Iona College. He is editor and chief of the International Journal of Intelligent Systems. He serves on the editorial board of a number of journals including the IEEE Transactions on Fuzzy Systems, Neural Networks, Data Mining and Knowledge Discovery, IEEE Intelligent Systems, Fuzzy Sets and Systems, the Journal of Approximate Reasoning and the International Journal of General Systems. In addition to his pioneering work in the area of fuzzy logic he has made fundamental contributions in decision making under uncertainty and the fusion of information.
[ Return to top ]
Michio Sugeno
Abstract The human brain consists of a neural system as hardware and a language system as software. It is, therefore, necessary to take two approaches to create the human brain. While the hardware-centered approach is based on computational neuroscience, it is possible to base the soft ware-centered approach on linguistics. With this in mind, we discuss the language functions in the brain. There are three approaches to explore the language functions: top-down, intermediate, and bottom-up. In top-down approach we start from existing phenomena of language and in bottom-up approach we start from neural processes to deal with language. Intermediate approach means something between the two. We adopt, as the basic theory, Systemic Functional Linguistics initiated by Halliday. In a top-down approach, we have developed a computational model of language which consists of the semiotic base describing the system of language, and text understanding/generation with the semiotic base. As to an intermediate approach, we discuss the stratified system of language in the brain by introducing some clinical evidence obtained from studies on aphasia. In a bottom-up approach, we have conducted brain experiments to analyze dynamical processes in understanding the meanings of texts with and without honorific expressions. Biography Michio Sugeno
received D. Eng. from Tokyo Institute of Technology, Tokyo, Japan.
Currently Dr. Sugeno is a distinguished visiting professor of Doshisha
University, Kyoto, Japan and a distinguished affiliated researcher of
European Centre for Soft Computing, Oviedo, Spain. Dr. Sugeno is the
past president of IFSA and IFSA Fellow. He received IEEE Pioneer Award
in Fuzzy Systems. His main research interests lie in language functions
in the brain with a perspective of Systemic Functional Linguistic, and
fuzzy measures /integrals for evaluation and decision making.
[ Return to top ]
Bernadette Bouchon-Meunier Similarities in Fuzzy Data Mining: From a Cognitive View to Real-World Applications Abstract Fuzzy logic provides interesting tools for data mining, mainly because of its ability to represent imperfect information, for instance by means of imprecise categories, measures of resemblance or aggregation methods. This ability is of crucial importance when databases are complex, large, and contain heterogeneous, imprecise, vague, uncertain, incomplete data. We focus our study on the use of similarities which are key concepts for all attempts to construct human-like automated systems or assistants to human task solving since they are very natural in the human process of categorization underlying many natural capabilities such as language understanding, pattern recognition or decision-making. We base our discourse on cognitive approaches of similarities, stemming for instance from Tversky's and Rosch's seminal works, among others. We point out a general framework for measures of comparison compatible with these cognitive foundations, and we show that measures of similarity can be involved in many steps of the process of data mining, such as clustering, construction of prototypes, utilization of expert or association rules, fuzzy querying, for instance. We eventually illustrate our discourse by examples of similarities used in real-world data mining problems. Biography Bernadette
Bouchon-Meunier is a director of research at the National Center for
Scientific Research, head of the department of Databases and Machine
Learning in the Computer Science Laboratory of the University Paris 6.
Graduate from the Ecole Normale Superieure at Cachan, she received the
degrees of B.S. in Mathematics and Computer Science, Ph.D. in Applied
Mathematics and D. Sc. in Computer Science from the University of Paris.
Editor-in-Chief of the International Journal of Uncertainty, Fuzziness
and Knowledge-based Systems (World Scientific), she is also a member of
the editorial board of the International Journal of Approximate
Reasoning, Fuzzy Sets and Systems, International Journal of Fuzzy
Systems, International Journal of Information Technology and Intelligent
Computing, Journal of Uncertain Systems. She is the (co)-editor of
twenty books and the (co)-author of four books in French on Fuzzy Logic
and Uncertainty Management in Artificial Intelligence. She is a
co-founder and co-executive director of the International Conference on
Information Processing and Management of Uncertainty in Knowledge-based
Systems (IPMU) held every other year since 1986. She is presently an
elected member of the Administrative Committee of the IEEE Computational
Intelligence Society and a member of the IEEE Women in Engineering
committee, chair of the IEEE French Chapter on Computational
Intelligence. She has also chaired the IEEE Women in Computational
Intelligence Committee from 2004 to 2007.
[ Return to top ]
Dario Floreano Evolution of Cooperation in Biological and Robotic Societies Abstract Cooperation is widely spread in nature and takes several forms, ranging from behavioral coordination to sacrifice of one's own life for the benefit of the society. The behavioral rules that lead to cooperation are interesting for robotics too when several robots are required to accomplish the same mission. However, the interactions among robots sharing the same environment can amplify in unexpected ways, or silence, the behavior of individual robots, making very difficult the design of rules that produce stable cooperative behavior. It is thus interesting to examine under which conditions stable cooperative behavior evolves in nature and how those conditions can be translated into evolutionary algorithms that are applicable to a wide range of robots. In this talk I will review biological theories of evolution of cooperative behavior and focus on the theories of kin selection and group selection. I will show how these two theories can be mapped into different evolutionary algorithms and compare their efficiency in producing control systems for a swarm of sugar-cube robots in a number of cooperative tasks that vary in the degree of requested cooperation. I will then describe an example where the most efficient algorithm is used to evolve a control system for a swarm of aerial robots that must establish a radio network between persons on the ground. In another set of experiments I describe how those evolutionary conditions can be tested for the emergence of communication where colonies of "expressive" robots are exposed to food and danger sources that cannot be uniquely be identified at distance. Here, communication of the source type brings an advantage to the colony at the expense of the individuals that decide to tell which is the food or poison. The results shed light on the conditions that may have favored the evolution of altruistic cooperation and communication. Biography Dario
Floreano is Associate Professor in the School of Engineering at the
Swiss Federal Institute of Technology in Lausanne (EPFL) where he is
director of the Laboratory of Intelligent Systems, of the Institute of
Systems Engineering, and responsible for the EPFL Master curriculum in
Robotics and Autonomous Systems. His research activities include
embodied cognitive science, evolutionary robotics, bio-mimetic robotics,
neural computation, and biology reverse engineering. Dario published
more than 100 peer-reviewed papers, authored 2 books, and edited 3 other
books. He co-organized 11 international conferences and joined the
program committee of approximately 100 conferences. He is on the
editorial board of 9 international journals: Neural Networks; Genetic
Programming and Evolvable Machines; Adaptive Behavior; Artificial Life;
Connection Science; Evolutionary Computation; IEEE Transactions on
Evolutionary Computation; Autonomous Robots; Evolutionary Intelligence
(Jan 2008). He is also editor-in-chief of the podcast "Talking Robots"
featuring interviews with key figures in Robotics and A.I. He is
co-founder and member of the Board of Directors of the International
Society for Artificial Life, Inc. and member of the Board of Governors
of the International Society for Neural Networks. Dario was involved in
the launch of several research programs by the European Commission in
the areas of Future Emergent Technologies, Robotics, Control, and
Complex Systems. [ Return to top ]
Hans-Paul Schwefel Simulated Evolution Under Multiple Criteria Conditions Revisited Abstract For sure, organic evolution is confronted with a couple of difficulties that are nightmares for traditional optimization algorithms. Among those difficulties are situations in which survival demands the ability to affront not only one peril, but several ones at the same time. More often than not some of the survival criteria are in conflict with each other. Evolutionary algorithms, which have entered successfully the market of solving difficult optimization problems, therefore have been extended to the more general class of multiple criteria optimization - not from the very beginning, but al least during the last decade. The first part of this article/talk tries to present an overview of different approaches that have been proposed, implemented, analyzed, and used in practice. The question is then raised, whether these approaches, though successful, reflect mechanisms that are found in nature, so that they can be called bio-inspired - or not. If not, the next question is, how organic evolution deals with multiple objectives, represented for example by different predator species or challenges like diseases and environmental stresses. Though no final answers can be presented, the attempt is made to highlight at least one direction of further efforts to create algorithms that both solve multicriteria problems effectively and deliver an explanation how organic evolution really works. Biography Hans-Paul Schwefel, born in December 1940 at Berlin, studied Aero- and Space-Technology at the Technical University of Berlin (TUB). Before and after receiving his engineer diploma in 1965, he worked at the Hermann-Foettinger-Institute of Hydrodynamics, from 1967 to 1970 at the industrial AEG research institute, and from 1971 to 1975 again at the TUB, from where he got his Dr.-Ing. degree in 1975. Coherent during that period at Berlin was the development of a new experimental and later on also numerical optimization method called Evolutionsstrategie. From 1976 to 1985 he acted as senior research fellow at the Research Centre (KFA) Jülich, where he was head of a computer aided planning tools group. Since 1985 until he was pensioned in 2006 he was holder of a Chair for Systems Analysis at the University of Dortmund, Department of Computer Science. From 1990 to 1992 he acted as dean of the faculty, from 1997 to 2004 as spokesman of the collaborative research center on computational intelligence (SFB 531), and from 1998 to 2000 as pro-rector for research and junior scientists at the university. He is member of the editorial boards of the journals Evolutionary Computation (MIT press) and Natural Computing (Kluwer/Springer) and advisory board member of the Springer book series on Natural Computation. He was elevated to Fellow of the IEEE in 2007. In 1990 he was co-founder of the international conference series on Parallel Problem Solving from Nature (PPSN), which has been held biennially ever since.
[ Return to top ]
David Wolfe Corne
Abstract In this talk
I will try to argue why single-objective optimization should be
outlawed, and then summarize and discuss some of the highlights and
current directions in multiobjective optimization research. Then, I will
try to predict what we will be doing when multiobjective optimization is
considered to be "the past". It will be interesting to anyone who, like
me, is enamoured with the concept of landscapes, but tends to get lost
in them. Biography David Corne
is a Professor of Computer Science at Heriot-Watt University, Edinburgh,
UK. He is head of the Intelligent Systems Lab, which works across the
scope of intelligent systems, with projects and major achievements in
each main area of computational intelligence. His own interests are
many, with a focus on large scale optimization, multiobjective
optimization, and applications in bioinformatics, medicine and
communications. He started out with degrees in mathematics and
artificial intelligence (respectively), and was a researcher in the
Department of Artificial Intelligence, University of Edinburgh for six
years, working first on intelligent design support systems (with Tim
Smithers), and then on evolutionary scheduling and timetabling (with
Peter Ross and Hsiao-Lan Fang), producing some early and influential
ideas and techniques which have since become common in applications. He
moved to the University of Reading in 1995, and built up a track record
in various aspects and applications of evolutionary computation,
notably, new algorithms and theory in evolutionary multiobjective
optimization (with Joshua Knowles). More recently he has developed novel
approaches to very-many-objective and large-scale optimization
[ Return to top ]
Garrison W. Greenwood
Abstract Self-adaptive systems autonomously change their behavior to compensate for faults or to improve their performance. Evolvable hardware, which combines evolutionary algorithms with reconfigurable hardware, is often proposed as the cornerstone for systems that use self-adaption for fault recovery. Although evolvable hardware was first introduced over 15 years ago, there are few, if any, fault tolerant self-adaptive systems in operation today. One primary reason why these unfortunate circumstances have arisen is many designers―and not limited to just designers from the computational intelligence community―do not really understand how to build a basic fault tolerant system, let alone a self-adaptive fault tolerant system. This talk describes how fault tolerant systems are built. Special accentuation is given to systems that use self-adaption as the fault recovery mechanism. The advantages and disadvantages of intrinsic evolvable hardware fault recovery methods are discussed and design guidelines are presented. Biography Garrison Greenwood received the Ph.D. degree in electrical engineering from the University of Washington. After spending more than a decade in industry designing multiprocessor embedded system hardware, he entered academia where he is now an associate professor in the Department of Electrical and Computer Engineering at Portland State University. In 1999 and 2000 he was a National Science Foundation Scholar-in-Residence at the National Institutes of Health in Bethesda, Maryland. Dr. Greenwood has served as a organizing committee member on many international conferences and was the general chair of the 2004 Congress on Evolutionary Computation. In 1999 he was an associate editor of the IEEE Transactions on Neural Networks, and since 2000 has been an associate editor of the IEEE Transactions on Evolutionary Computation. He is currently serving a second two-year term as Vice-President of Conferences for the IEEE Computational Intelligence Society. He is a member of Tau Beta Pi, Eta Kappa Nu, is a senior member of the IEEE and is a registered professional engineer in the State of California. His research interests are evolvable hardware, adaptive systems, and game theory.
[ Return to top ]
Kay Chen Tan
Abstract Evolutionary algorithms are stochastic search methods that are efficient and effective for solving sophisticated multi-objective (MO) problems. Advances made in the field of evolutionary multi-objective optimization (EMO) are the results of two decades worth of intense research, studying various topics that are unique to MO optimization. However many of these studies assume that the problem is deterministic, and the EMO performance generally deteriorates in the presence of uncertainties. In certain situations, the solutions found may not even be implementable in practice. In this talk, the challenges faced in handling three different forms of uncertainties in EMO will be discussed, including 1) noisy objective functions, 2) dynamic MO fitness landscape, and 3) robust MO optimization. Specifically, the impact of these uncertainties on MO optimization will be described and the approaches/modifications to basic algorithm design for better and robust EMO performance will be presented. Biography Kay Chen TAN
is currently an Associate Professor in the Department of Electrical and
Computer Engineering at the National University of Singapore, Singapore.
He is actively pursuing research in the field of computational
intelligence, with applications to multi-objective optimization,
scheduling, design automation, and games. Dr Tan has published over 70
journal papers, 100 papers in conference proceedings, and co-authored 5
books including Multiobjective Evolutionary Algorithms and Applications
(Springer-Verlag, 2005), Modern Industrial Automation Software Design
(John Wiley, 2006), Evolutionary Robotics: From Algorithms to
Implementations (World Scientific, 2006), Neural Networks: Computational
Models and Applications (Springer-Verlag, 2007), and Evolutionary
Multi-objective Optimization in Uncertain Environments: Issues and
Algorithms (Springer-Verlag, expected in 2008). Dr Tan has been invited
to be a keynote/invited speaker for many international conferences. He
also served in the international program committee for over 60
conferences and involved in the organizing committee for over 20
international conferences, including the General Co-Chair for IEEE
Congress on Evolutionary Computation 2007 in Singapore and the General
Co-Chair for IEEE Symposium on Computational Intelligence in Scheduling
2007 in Hawaii. Dr Tan is currently a member of Board of Directors in
Evolutionary Programming Society, USA. Dr Tan currently serves as an
Associate Editor / Editorial Board member of 8 international journals,
including IEEE Transactions on Evolutionary Computation, Journal of
Scheduling, European Journal of Operational Research, and International
Journal of Systems Science. Dr Tan was a winner of the NUS Outstanding
Educator Awards (2004), the NUS Engineering Educator Awards (2002, 2003,
2005), the NUS Annual Teaching Excellence Awards (2002, 2003, 2004,
2005, 2006), and the NUS Teaching Awards Honour Roll (2007).
|
