摘要
针对遗传算法、粒子群算法等BP网络的学习算法对高维复杂问题仍易早熟收敛,且无法保证收敛到最优解。把量子粒子群算法应用于BP网络的学习中,并把改进BP网络用于入侵检测。通过KDD99CUP数据集分别对基于不同学习算法的BP网络进行了实验比较,结果表明:该算法的收敛速度较快,可在一定程度上提高入侵检测系统的准确率和降低的误报率。
Aimed at the problem of BP networks learning algorithm such as genetic algorithms, particle swarm optimization algorithm is still easy to premature convergence, and can not guarantee convergence to the optimal solution for high-dimensional complex issues. The quantum particle swarm algorithm is applied to BP networks and improve the BP networks for intrusion detection. By KDD99 CUP data set, experiments of BP networks based on different learning algorithm were compared. Results show that convergence speed of the algorithm is fast and can improve accuracy of intrusion detection systems and reduce the false alarm rate to some extent.
出处
《传感器与微系统》
CSCD
北大核心
2010年第2期108-110,共3页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(60703035)
重庆市自然科学基金资助项目(2009BB2288)
关键词
网络安全
入侵检测
BP网络
量子粒子群
network security
intrusion detection
BP networks
quantum particle swarm
