Department Seminar - by Dr. Yixin Chen (Univ. of Mississippi)

Many computational methods have been developed to analyze data from different viewpoints, e.g., prediction, feature subset selection, and data interpretation. As these methods are geared towards specific purposes, their domains of application usually do not overlap: classifiers with high prediction performance , e.g., support vector machines (SVM) or random forest (RF), may be a "black box" model lacking interpretability; while an easy-to-interpret model, such as a decision tree, does not compete well with SVM or RF on classification accuracy. In this talk, we will discuss recent progress we made along the direction of the improving interpretability of prediction models. Specifically, we will discuss our recent work on feature subset selection, problem space partition, and rule-based learning.

Yixin Chen received B.S.degree (1995) from the Department of Automation, Beijing Polytechnic University, the M.S. degree (1998) in control theory and application from Tsinghua University, and the M.S. (1999) and Ph.D. (2001) degrees in electrical engineering from the University of Wyoming. In 2003, he received the Ph.D. degree in computer science from The Pennsylvania State University. He had been an Assistant Professor of computer science at University of New Orleans. He is now an Associate Professor at the Department of Computer and Information Science, the University of Mississippi. He is the recipient of 2011 University of Mississippi, School of Engineering Junior Faculty Research Award and the recipient of 2012 University of Mississippi, School of Engineering's Outstanding Faculty Member of the Year. His research interests include machine learning, data mining, computer vision, bioinformatics, and robotics and control.