基于神经网络与遗传算法的入侵检测研究

本文针对入侵检测系统中对入侵事件的误报和漏报,提出了一种将神经网络和遗传算法相结合的方法。该方法基于神经网络算法的局部精确搜索和遗传算法的全局搜索特性,用遗传算法优化神经网络权值,既克服了神经网络算法易陷入局部极值的弊端,又解决了单独使用遗传算法在短时间内难以找到最优解的问题。将得到的网络结构用于入侵检测系统中,使之能够准确的找出已知的攻击行为,并能够发现新的攻击行为。仿真结果表明该方法具有一定的有效性。

Research of intrusion detection based on the neural networks and genetic algorithm

This paper shows a way that combines neural network with genetic algorithm aiming at the distorts and missing of intrusions in intrusion detection system(IDS).This method optimizes the weights of the neural network utilizing the local precise search of neural network algorithm and global search characteristic of genetic algorithm.Not only conquers it the abuse of neural network algorithm that easily trapped into local extremum,it but also solves the problem of genetic algorithm that hardly gaining the optimal solution in a short time.And the network structure is applied into intrusion detection systems,to find out the known detection exactly,and the new detection.The experiment shows its validity.