一种改进免疫遗传算法及在异常检测中的应用

An Improved Immune Genetic Algorithm and Its Application in Anomaly Detection

该文针对免疫遗传算法的不足,在分析其特性的基础上,引入了隔离小生境技术,改进交叉算子和变异算子,提出一种改进算法。在基于模糊关联规则挖掘的异常检测中采用本算法优化后的隶属函数,能够扩大正常关联规则集之间的相似度,缩小正常与异常关联规则集之间的相似度,提高异常检测的性能。通过以网络流量为数据的异常检测实验仿真对算法进行了验证。实验结果说明了该算法的可行性和有效性。

In view of immune genetic algorithm＇s shortages, an improved algorithm was introduced. The proposed algorithm used isolated niche technology, improved the cross and mutation operation based on immune genetic algo-rithm. The optimized membership functions were used in fuzzy association rules mining to anomaly detection. It could magnify the similarity between normal association role sets, and reduce the similarity between a normal and an abnormal association rule set at the same time. So it could improve the performance of anomaly detection. Feasibility of the algorithm was verified by experiments on anomaly detection based on network traffic.