A hybrid genetic algorithm for designing feedforward neural networks

In this paper, a hybrid algorithm is proposed for designing feedforward neural networks. A genetic algorithm is proposed to tune the connections and parameters between the input layer and the hidden layer, and orthogonal transformation is applied to tune the connections and parameters between the hidden layer and the output layer. The crossover operator and mutation operator are based on the singular value decomposition of the outputs of the hidden nodes. Using the proposed algorithm, both the structure and parameters of a neural network can be optimized efficiently. Simulations are presented to demonstrate the effectiveness of the proposed approach.