Lipo Wang, Nanyang Technological University, Singapore Title: Expectation-Maximization based Rough Set Feature Selection

Abstract: We use Rough Set theory for feature selection based on tolerance classes that extend the existing equivalent classes. Given a discrete dataset, it is possible to find the equivalent classes; however, in case of datasets with real-valued attributes, it is impossible to say whether two objects are the same, or to what extent they are the same, using the indiscernibility relation. Tolerance-based Rough Set extended the RS theory using the tolerance or similarity relation to find the similar objects. The determination of initial tolerance classes is a challenging and important task for accurate feature selection and classification. Here, the Expectation- Maximization clustering algorithm is applied to determine initial tolerance classes for real-valued datasets. This method generates significantly fewer features with either a higher or the same accuracy compared with two existing methods, i.e., fuzzy lower approximation-based FS and fuzzy boundary region-based FS, on a number of benchmarks from the UCI repository.

Bio Sketch: My research interest includes neural networks, fuzzy systems, and chaos, with applications to data mining, bioinformatics, multimedia, and optimization. I am author or co-author of over 75 journal publications, 15 book chapters, and 110 conference presentations. I hold a U.S. patent in neural networks. I have authored 2 monographs and edited 15 books. I was keynote/panel speaker for several international conferences. I have been Associate Editor of IEEE Transactions on Knowledge and Data Engineering, IEEE Transactions on Neural Networks and IEEE Transactions on Evolutionary Computation, as well as Area Editor of the Soft Computing journal and Editorial Board Member of 19 additional international journals. I served as Vice President for Technical Activities (2006-2007) and Chair of Emergent Technologies Technical Committee (2004-2005), IEEE Computational Intelligence Society. I was President of the Asia-Pacific Neural Network Assembly (APNNA) and received the 2007 APNNA Excellent Service Award. I was Founding Chair of both the IEEE Engineering in Medicine and Biology Chapter Singapore and IEEE Computational Intelligence Chapter Singapore. I serve/served on the steering/advisory/organizing/program committees of over 160 international conferences.