A Rough CP Neural Network Model Based on Rough Set

Aiming at the problem that the counter propagation network (CPN) can not make good use of nerve cells, a rough CP neural network model based on rough set is proposed. It changes the strategy of winning as the king, and decides output values to use Rough Member Function, which expresses the level of an element subordinated to a set. Experiments show that the approach can solve some problems in other neural Networks, for example, sample size and quality. They would directly influence the accuracy. While reducing training time, the prediction precision of the network can be greatly improved.