Multi-agent evolutionary design of Beta fuzzy systems

This paper provides an overview on a new evolutionary approach based on an intelligent multi-agent architecture to design Beta fuzzy systems (BFSs). The Methodology consists of two processes, a learning process using a clustering technique for the automated design of an initial Beta fuzzy system, and a multi-agent tuning process based on Particle Swarm Optimization algorithm to deal with the optimization of membership functions parameters and rule base. In this approach, dynamic agents use communication and interaction concepts to generate high-performance fuzzy systems. Experiments on several data sets were performed to show the effectiveness of the proposed method in terms of accuracy and convergence speed.