Abstract: Intrusion detection is a central issue of network security research. This paper proposes a new neural network ensembles model for intrusion detection system. The model trains the individual networks based on data reducing. Genetic algorithm is used to optimize neural network weight. Neural network techniques are used to combine the different classification results. Theory and experiment show that the model is effective.

Key words: neural network ensembles, attribute selection, genetic algorithm, intrusion detection

摘要： 入侵检测是网络安全研究中的热点。提出了一种用于入侵检测的神经网络集成模型。该模型采用神经网络集成分类技术，去除训练集中的冗余数据，利用遗传算法优化成员网络的权值，在此基础上训练成员网络，最终通过神经网络对成员网络的输出结果进行融合。理论和实验表明，模型具有较好的检测能力。

关键词: 神经网络集成, 属性选择, 遗传算法, 入侵检测