Parameters Optimization of Back Propagation Neural Network Based on Memetic Algorithm Coupled with Genetic Algorithm

Memetic algorithm is both global and local search algorithm based on evolutionary algorithm; indeed, it has high optimization efficiency and fast searching speed. The paper proposes an efficient method of choosing the proper parameters to set up the moderate-scaled and efficient back propagation neural network via memetic algorithm. In addition, weights and thresholds of BPNN are also optimized by means of the Genetic algorithm in order to build up BPNN with high performances; such as simplified structure, low store memory occupation, high prediction accuracy and generalization etc. Experiments perform on the Wisconsin breast cancer diagnosis datasets from UCI Machine Learning Repository, which show through a serial of simulation results that the proposed algorithm has better accuracy than the state-of-the-art algorithm.