Research on anti-glycation activity based on dynamic particle swarm optimization for BP neural network

摘要

Plenty of pharmacological and clinical experiments have proved that polysaccharide has high pharmaceutical value, as mainly demonstrated in the fact that polysaccharide can improve the immune function, anti-tumor, anti-viral, anti-aging, anti-diabetes and anti-radiation of organisms. This paper is mainly about the research on anti-glycation activity based on dynamic particle swarm optimization (DPSO) for BP neural network. BP neural network has been widely used in every field, including bio-medicine. As a non-linear artificial intelligent system, it can look for the complex correlation among variables, recognize and build a model for the input variables, and output the direct non-linear relationship. This paper combines PSO with BP neural network for the network training and prediction research of the anti-glycation activity data in biomedicine. The prediction based on artificial neural network has been gradually applied in the research of biomedicine and the topological structure of its model includes the input layer, the hide layer and the output layer. When the actual output is inconsistent with the expected output, it enters into the back propagation phase of errors. The error passes the output layer, corrects the weights of every layer in the same way of error gradient descent and starts back propagation to the hide layer and the input layer. This process continues until the error output by the network is acceptable or reaches the pre-set number of learning. The experimental results show the proposed method has satisfactory results, better convergence, and improves the prediction accuracy.