摘要
针对BP算法及其改进算法泛化能力不强的问题,探讨用贝叶斯正规化算法来提高网络泛化能力。结果表明在相同网络规模或误差条件下,贝叶斯正规化算法泛化能力明显优于基本BP算法及其它改进的BP算法,且收敛速度较快,拟合效果较好。
To counter the problem that the general BP algorithms and its improvements have weak generalization capacity, we study the Bayesian Regularization algorithm to enhance the neural network generalization capacity. Based on the same network size and error probability, the results show that Bayesian Regularization algorithm has better generalization capacity than, the basic BP algorithms, early stopping and other improved BP algorithms, and furthermore, it has higher convergence speed and better effects of approximation.
出处
《数理医药学杂志》
2007年第3期293-295,共3页
Journal of Mathematical Medicine
关键词
BP神经网络
贝叶斯正规化
泛化能力
BP neural network
bayesian-regularization
generalization capacity
