A Fuzzy Classifier Based on Imperialist Competitive Algorithm

Fuzzy classifiers as a kind of fuzzy systems are powerful approaches in pattern recognition tasks. Each fuzzy classifier includes important parameters which play effectiveroles on its performance. Type and locations of membership functions, in addition to optimum number of fuzzy rulescontaining suitable antecedents and consequents are most important of these structural parameters. Usually, the major problem in design and implementation of fuzzy classifiers isoptimum setting up of these parameters, to reach the best performance. In this paper, membership function shapes andthe fuzzy rules of fuzzy systems are adjusted by using a Imperialist Competitive Algorithm (ICA) in order to obtain an optimal fuzzy system. Extensive experimental results are presented to show the effectiveness and powerfulness of the proposed method

کلیدواژه ها:

Imperialist Competitive Algorithm ، fuzzy system ، classifier