Genetic design for feedforward neural network

During application of neural network, there are some problems, including difficult determination of the size and structure of neural network in advance, the slow learning speed of neural network, and easy to converge to local optimum. In the view of these problems, Genetic Algorithms (GAs) is recommended to design and train neural network, the top structure and related parameters (weights and thresholds) can be obtained simultaneously. On this basis, some improvement on proposed method were performed, which mainly includes that float-point matrix is adopted to encoding, evolution of GAs itself is modified and the proposed method can satisfy some constrained conditions. For feedforward neural network, this method can find suitable network structure and corresponding parameter (weights and thresholds) simultaneously under certain constrained conditions. New method has great improvement over the old one in both accuracy and speed.