主題演講

題目: Computational Intelligent Brain Computer Interaction and Its Applications on Driving Cognition

演講者:

林進燈 講座教授

學經歷:

國立交通大學電機與控制工程學系講座教授
美國普渡大學電機博士
國科會97年度傑出特約研究員獎

專長:

Computer Vision, Fuzzy Neural Network, Speech Processing and Control, Brain-Computer Interface

摘要:

Human cognitive functions such as perception, attention, memory and decision making are omnipresent in our daily life activities. For instance, driving is one of the most common attention-demanding tasks in our daily routine. Driver’s fatigue, drowsiness, distraction or motion sickness is reported as the some of major causal factors in many traffic accidents. When drivers lost their attention, they had appreciably reduced (or diminished) the perception, recognition and vehicle control abilities. Based on these causalities, how to effectively prevent and enhance the human cognitive functions has become a very important issue. Recently, many investigators had developed novel algorithms based on computational intelligence (CI) technologies such as Fuzzy Logic and Fuzzy Neural Systems to monitor, maintain, or track the human operating performance. In this lecture, we briefly introduce the fundamental physiological changes of the human cognitive functions in driving first and then explain how to utilize these main findings to develop the monitoring and feedback systems based on Fuzzy logic and Fuzzy Neural technologies in the following two topics: (1) EEG-based cognitive state monitoring and prediction by using the self-constructing fuzzy neural systems; and (2) Spatial and temporal physiological changes and estimation of motion sickness. These research advancements can provide us new insights into the understanding of complex cognitive functions and lead to novel application enhancing our productivity and performance in face of real-world complications. The achievements are a natural follow up to the special issue on Brain-Computer Interactions, recently published in the November 2009 issue of IEEE Computational Intelligence Magazine.



題目: Recent Studies on Algorithms for Fuzzy Clustering

演講者:

Sadaaki Miyamoto

學經歷:

Professor, University of Tsukuba, Japan
1978, Ph.D., Applied Mathematics, Kyoto University, Japan
2007, Fellow, International Fuzzy Systems Association (IFSA)
2009, Fellow, International Society of Management Engineering (ISME)

專長:

Fuzzy Logic, Cluster Analysis, Data Mining

摘要:

Cluster analysis or data clustering has long been studied but recently many more researchers are interested in this area of studies, as data mining techniques are recognized to be very important in a variety of sciences and engineering. In this talk we overview recent studies on new methods and algorithms for fuzzy clustering, by focusing upon their theoretical aspects. We first note that fuzzy clustering can be divided into two categories of hierarchical fuzzy clustering and nonhierarchical fuzzy clustering. The former theory was established in 1990, when an old method of the single linkage in agglomerative hierarchical clustering is proved to be equivalent to the transitive closure of a symmetric fuzzy relation. Moreover a key concept for the equivalence is connected components of a fuzzy graph. After briefly noting this, we move to the discussion of the best-known method of fuzzy c-means clustering. We note there are two major objective functions of fuzzy c-means clustering that use a basic alternative optimization procedure with respect to cluster centers and membership matrix. The well-known objective function has been proposed by Dunn and Bezdek, while the other uses an additional term of entropy. The latter has been proposed by a number of researchers. We show these two methods have different theoretical properties by using fuzzy classifiers naturally derived from the two membership matrices. Although the method of entropy is less-known, we emphasize its importance in theoretical and methodological sense. Next, two recent studies on clustering are overviewed. First is clustering using kernel functions that are used in support vector machines. How kernel functions are used in fuzzy c-means clustering and related methods is described. Moreover cluster validity functions with and without kernel functions are shown and a simulation study comparing the effectiveness of different validity functions using many numerical examples is mentioned.