本文将多克隆算子和模糊C-均值算法相结合,将多克隆的免疫基因策略作用于特定的亲合度函数来指导聚类过程,得到一种无监督的聚类分析方法。该方法不依赖先验知识、与数据分布无关、能够有效处理大规模原始数据。在KDD CUP99数据集上进行了仿真实验,实验结果表明基于本文方法的入侵检测系统具备良好的性能。%A novel clustering algorithm is developed by using polyclonal algorithm and fuzzy C-means algorithm, which is directed by affinity function applied by polyclonal strategy. Independent of transcendental knowledge and data distribution, this algorithm can deal with mass unlabeled data, and the computer simulations on the KDD CUP99 dataset show that this method can achieve good performance.
展开▼