Title Statistical Tests for Computational Intelligence Research and Human Subjective Tests
Speaker Dr. Hideyuki Takagi
Chair Sanghamitra Bandyopadhyay

Abstract
When we compare convergence speed of multiple methods, for example, conclusions without statistical supports cannot be accepted. However, it is true that there are several students and researchers whose good research was not accepted due to wrong or no supports of statistical tests for their conclusions.

In this tutorial, we explain how to use what kind of statistical tests for which cases with exercise. Statistical tests that we study here include standard t-test, analysis of variance, sign test, Wilcoxon’s signed-ranks test, multiple comparisons, and others.

One of features of this tutorial is to show how to handle human subjective tests that cannot be avoided for interactive evolutionary computation (IEC) research and other human-related applications. We study Sheffe’s method of paired comparison that is one kind of ANOVA for this purpose.

Besides these basic lecture and exercise, we learn IEC applications and how statistical tests support the IEC research.

Biography
Hideyuki Takagi is received a doctorate in Engineering from Toyohashi University of Technology in 1991. He had worked for the Central Research Laboratories of Matsushita Electric Industrial Co., Ltd. (current Panasonic) in 1981 – 1995 and became Associate Professor of Kyushu Institute of Design and now Kyushu University after these two universities merged in 2003. He was one of trigger researchers of neuro-fuzzy booming started from 1988 from Japan and then to the world. Now, he is focusing on humanized computational intelligence, especially IEC research. He is a volunteer for IEEE SMC Society since 1998 as the Vice-President, Board of Governor, Technical Committee (TC) Chair, TC Coordinator, Distinguish Lecturer, Associate Editor of IEEE Trams. on SMC-B and now Trans. on Cybernetic, and one of organizers of SMC annual conferences. See his detail bio at http://www.design.kyushu-u.ac.jp/~takagi/