Abstract: According to that an intrusion detection system needs to achieve the best trade-off between detection rate and false positive rate，an improved multi-objective evolutionary algorithm is proposed to reduce the feature space and then select the best feature subset.The experiment results show that the best trade-off between detection rate and false positive rate can be achieved by the improved multi-objective evolutionary algorithm.The algorithm can improve the performance of the intrusion detection system.

Key words: intrusion detection, feature selection, multi-objective evolutionary algorithm, sequential search

摘要： 针对入侵检测系统要求检测率和误报率均衡优化，提出一种由顺序搜索策略改进的多目标进化算法，对特征空间进行压缩，以选择最优特征子集。实验结果表明，改进的多目标进化算法实现了检测率与误报率的均衡优化，较好地提高了入侵检测系统的性能。

关键词: 入侵检测, 特征选择, 多目标进化算法, 顺序搜索