PSOBPlunwen

  基于PSO的BP训练算法论文：在BP训练算法中，关于变权值、学习速率、步长的问题已被广泛地研究，几种基于启发式改进的技术也表明具有改善训练时间以及避免陷入局部最小的明显效果。这里BP训练过程由基于PSO同时优化log—Sigmoid函数与网络权值的新算 法(PSO。GainBP)实现。实验结果表明，PSO—GainBP比传统基于PSO的BP算法在网络训练方面具有更好的性能。(PSO based on the BP training algorithm Papers: In the BP training algorithm, on the right to change the value of learning rate, step length of the question has been widely studied, several heuristics based on improved technology also shows that to improve the training time and avoid falling into local the smallest tangible results. Here BP training process at the same time by the PSO-based optimization of log-Sigmoid function and the network weights of the new algorithm (PSO. GainBP) to achieve. The experimental results show that, PSO-GainBP than the traditional PSO based on the BP algorithm in network training has a better performance.)