【摘要】 为了克服BP神经网络速度慢、易陷入局部最小的缺点,利用GA的全局搜索能力优化BP神经网络权值,本文提出了遗传BP神经网络算法,并将其用于异常检测之中。在对Kddcup’99攻击数据进行分析和特征约简的基础上,设定了遗传BP神经网络算法的参数。实验结果表明,基于遗传BP神经网络异常检测模型的建立快于BP神经网络算法。更多还原

【Abstract】 In order to avoid the shortcomings of back-propagation neural networks which converge slowly and goes easily into the local minimum,the paper uses the global search capabilities of genetic algorithms to optimize the back-propagation neural network weights,proposes a genetic back-propagation neural network algorithm,and applies it to anomaly detection.Based on the analysis the KDDCUP’99 attack data and reduction features,the parameters of the algorithm are set.The experimental results show the proposed algorithm can study the anomaly detection model more quickly than the back-propagation neural network algorithm.更多还原