The Jisan Research Institute

Home • About Us • Research • Students • Student Programs • News • Contact Us

Introduction

This research group was put together as part of an NSF-Funded collaboration between the
Integrated Media Systems Center and the Jisan Research Institute. This project involved nine students and three research mentors. During the project, students learned topics relevant to the study of fuzzy systems and applied these, genetic algorithms, and a locally-developed metric designed to reduce the complexity of fuzzy systems. The students' system was able to accurately identify important rules and reduce the remaining rules, reducing the number of rules by 70-90% while still preserving the functionality.

Abstract

We explore the reduction of a fuzzy classifier designed to perform a binary classification of tracked or wheeled vehicles based on acoustic data. A genetic algorithm is used to explore the design space of the classifier, with variations performed on the number of antecedents included in the final fuzzy system. We discover that systems with reductions in the number of antecedents of between 50 and 95% perform well on this classification. A novel method of extracting important system components, known as \emph{open product analysis}, is applied, yielding systems that perform well with small numbers of antecedents. The fuzzy classifier we reduced performs well using only 10 to 15% of the antecedents that were originally used for classification.

Publications

T. Shin, D. Jue, D. Chandramohan, D. Choi, C. Seng, J. Yang, A. Bae, A. Lee, J. Lee, P. Lim, S. Kazadi, J. Mendel. Reduction of Fuzzy Systems through Open Product Analysis of Genetic Algorithm-Generated Fuzzy Rule Sets, Proceedings of the IEEE Conference on Fuzzy Systems, 2004, Budapest, Hungary.(postscript) (PDF)