利用聚类改进动态克隆选择算法的自体纯净性问题

动态克隆选择算法应用于入侵检测的过程中，经过记忆检测器和成熟检测器检测后的剩余抗原被直接作为自体供未成熟检测器耐受，但这些剩余抗原并非完全是自体，有可能隐含新型攻击。为此提出利用聚类分析技术进行改进，先用聚类算法将剩余抗原分成大、小簇，然后分析小簇中的数据，发现其中隐含的新型攻击，并及时更新记忆检测器集和自体集。实验结果表明，加入聚类分析的动态克隆选择算法能够增强检测系统发现未知入侵的能力。

Using Clustering to Improve Self-Purity of Dynamic Clonal Selection Algorithm

In the intrusion detection process of dynamic clonal selection algorithm, the antigens detected by memory detectors and maturity detectors are directly considered as self immature detectors to be tolerated. But there may be new attacks hidden in these antigens. To solve this problem, a new idea with clustering analysis is proposed. The clustering algorithm cluster remaining antigens then analyzes data existing in small cluster, finds hidden attacks and update memory detector set in time. The experimental results show that the dynamic clonal selection algorithm with clustering analysis can enhance the detection system's ability to discover unknown intrusions.