Abstract

In this paper, the problem of Type-2 Fuzzy Inference System’s optimization using Multi-Objective Particle Swarm Optimization (MOPSO) algorithm is presented. In order to optimize the type-2 fuzzy inference systems, we use the type-2 fuzzy weights of back-propagation neural networks. Comparing the results of the neural networks with type-2 fuzzy weights without optimization of the type-2 fuzzy inference systems, the neural networks with optimized type-2 fuzzy weights using the MOPSO algorithm, and the neural networks with type-2 fuzzy weights, we show that the bio-inspired methods have the superior performance. The comparative study is based on the Mackey-Glass time series problem for  = 17.

Keywords: type-2 fuzzy, neural network, time series, MOPSO algorithm.