摘要
首先将网络状态特征参数作为神经网络算法的个体,检测正确率作为适应度函数。通过遗传算法的选择、交叉和变异等操作对网络状态特征参数进行优化,引入一种新型的能量函数形式以解决收敛速度慢、目标函数局部极小问题。实验表明,该方法能够快速获得最优网络状态特征和分类器参数,提高网络入侵检测正确率。
The network state parameters are taken as the cell of the neural network algorithm, and the detection accuracy as fitness function. Through genetic algorithm selection, crossover and mutation operation, we introduce a new energy function to improve the convergence speed and solve the local minimum problem of the object function. Experiments show that the method can optimize the neural network state parameters and classifier parameters, and upgrade the detection accuracy rate.
出处
《长春工业大学学报》
CAS
2012年第4期447-450,共4页
Journal of Changchun University of Technology
关键词
网络入侵
特征选择
神经网络
network intrusion
feature selection
neural network.
