Designing Neural Networks Using Novel Water Wave Optimization Based Memetic Algorithm

Searching neural networks' (NN) optimal parameters can be seen as a multi-modal optimization problem. This paper proposes a memetic water wave optimization algorithm (WWOMA) to determine the optimal weights of NN. In the proposed WWO-based MA (WWOMA), we employ WWO to perform global search by both individual improvement and population co-evolution, and then employ several local search components to enhance its local refinement ability. Furthermore, a Meta-Lamarckian learning strategy is utilized to choose a proper local refinement component to concentrate computational efforts on more promising solutions. We carry out numerical experiments on six well-known NN designing benchmark problems, the empirical results demonstrate both the feasibility and effectiveness by applying WWOMA to design NN.