摘要： 针对传统遗传算法在网络入侵检测中存在分类复杂的问题，提出结合条件熵遗传算法(CEGA)和支持向量机(SVM)的网络入侵检测算法。将入侵特征的抽取和分类模型的建立进行联合优化，同时利用训练数据的统计特性指导入侵特征的抽取，并对特征空间进行线性变换，得到优化的特征子集和分类模型，在提高分类检测率的同时降低检测时延。

关键词: 入侵检测, 遗传算法, 支持向量机

Abstract: Due to the complex classification problems from the traditional genetic algorithm in the process of network intrusion detection , this paper proposes the network intrusion detection algorithm combined Conditional Entropy Genetic Algorithm(CEGA) and the Support Vector Machine (SVM). To optimize jointly the invasion of feature extraction and classification model, while taking advantage of the statistical characteristics of training data to guide the invasion feature extraction, and according to the feature space linear transformation to obtain the optimal feature subset and the classification model, as improving the classification test rates, the detection latency is reduced.

Key words: intrusion detection, genetic algorithm, Support Vector Machine(SVM)

中图分类号: 

• TP391