关键词: BP网络, 权值调整, 参数优化

Abstract: A new BP algorithm optimized by ant colony algorithm was proposed. Training error and test error of BP network were used to update and choose the information, in order to calculate the transfer probability in ants' route. Parameters value of sites in ants' route was bestowed on BP network, while parameters and training error stored in stored units were changed along with the adjustment of training error of BP network. Through iterative and mutual optimization, the best-optimized parameters of BP network were obtained. BP network based on ant colony algorithm was validated and tested for optimizing several functions. The results show that the training number of optimized BP network is smaller, and the precision of model is higher compared with the traditional BP algorithm while the training errors are of the same order of magnitude.

Key words: BP Neural Network, weight adjustment, parameters optimization