The neural network based on rough set theory

In order to simplify the complexity of the BP network structure and reduce the time of training samples. The article simplifies the complexity of BP network structure through the research of the BP network and rough set, removes the samples´ redundant attribute using the rough set attribute reduction theory, and trades the reduction after the BP network data as training samples; It also reduces the time of training samples by training the samples with the Conjugate Gradient Algorithm with Momentum and Batch Techniques. The experimental results show the effectiveness of the method.