改进蚁群算法结合BP网络用于入侵检测

Improved ant colony algorithm and BP network for intrusion detection

针对用BP神经网络进行入侵检测时权值难以确定的问题,提出一种基于改进蚁群算法与BP网络的入侵检测方法。基于蚁群算法构建解特点,正反馈自催化机制和分布式计算机制和BP网络局部精确搜索的特性,将蚁群算法和BP算法有机结合,利用蚁群算法优化BP网络,并对蚁群算法进行改进。通过KDD99CUP数据集分别对基于不同算法集合的BP神经网络进行了仿真实验,结果表明:改进算法收敛速度快,迭代次数较少,可在一定程度上提高入侵检测系统的准确率。

This paper presents an intrusion detection method based on an improved ant colony algorithm and an optimized BP network. According to the construction characteristics of an ant colony algorithm,and the characteristics of positive feedback from the catalytic mechanism,the distributed computer system and the local search of BP network,an organic combination of BP algorithm and an ant colony algorithm is achieved to optimize the BP network using an ant colony optimization algorithm. In turn the ant colony algorithm is also improved. Using KDD99 CUP data set,the BP neural networks based on different algorithms are simulated. The simulation results show that the faster convergence of the algorithm,the less number of iterations,and the improvement on the accuracy of intrusion detection systems.