Cooperative Particle Swarm Optimization for TSK-Type Neural Fuzzy Systems

This study proposes a cooperative particle swarm optimization (CPSO) to optimize the parameters of the TSK-type neural fuzzy system (TNFS) for classification applications. The proposed CPSO uses cooperative behavior among multiple subswarms to decompose the neural fuzzy systems into rule-based subswarms, and each particle within each subswarm evolves by a specific particle swarm optimization (PSO) separately. Therefore, the CPSO can accelerate the search and increase global search capacity. Finally, the TNFS with CPSO (TNFS-CPSO) is adopted in several classification applications. Experimental results demonstrate that the proposed TNFS-CPSO method has a higher accuracy rate and a faster convergence rate than the other methods.