Improved Adaptive Neuro-Fuzzy Inference System with Imperialist Competitive Learning Algorithm (ICA-ANFIS)

This paper introduces a new type of adaptive Neuro-fuzzy inference system, denoted as ICA-ANFIS (Adaptive Neuro-fuzzy Inference System with Imperialist Competitive Learning Algorithm). The previous learning algorithms of ANFIS emphasized on gradient based methods or least squares (LS) based methods, but gradient computations are very computationally and difficult in each stage, also gradient based algorithms may be trapped into local optimum. This paper introduces a new hybrid learning algorithm based on imperialist competitive algorithm (ICA) for training the antecedent part and least square estimation (LSE) method for optimizing the conclusion part of ANFIS. This hybrid method is free of derivation and solves the trouble of falling in a local optimum in the gradient based algorithm for training the antecedent part. The proposed ICA-ANFIS system is applied for prediction of Mackey-Glass chaotic time series. Analysis of the obtained results and comparisons with recent and old studies demonstrates the promising performance of this new approach.

Keywords:

Gradient based , Imperialist competitive algorithm , Fuzzy systems , Fuzzy neural networks , ANFIS , least square estimation