Tuning of the Structure and Parameters of a Neural Network Using a Hybrid Good Point Set Evolutionary Strategy

In this paper, a hybrid good point set-evolutionary strategy is applied to solve the problem of tuning both network structure and parameters of a feedforward neural network. PSO frame can make the resulting evolutionary algorithm more robust and statically sound, especially for global optimization. Good Point Set can make the local search achieve the same sound results just as the state-of-the-art methods do, such as orthogonal method. But the precision of the algorithm is not confined by the dimension of the space. An integrated mechanism is used to enrich the exploration and exploitation abilities of the approach proposed.The presented approach is effectively applied to solve the examples on forecasting the sunspot numbers.