摘要： 针对直觉模糊c-均值(IFCM)聚类算法易陷入局部最优的问题，从适应度值标定和群体多样化2个方面对遗传算法(GA)进行优化，并将优化后的GA与IFCM算法相结合，提出一种改进的IFCM算法用于入侵检测。优化后的GA具有更优良的全局寻优特性，与IFCM算法结合后，可避免算法陷入局部最优。在KDD CUP99数据集上的仿真结果表明，与IFCM算法相比，改进算法能有效提高聚类精度和检测效率。
关键词:
直觉模糊c-均值,
聚类,
局部最优值,
遗传算法,
全局寻优,
入侵检测
Abstract: Concerning that the Intuitionistic Fuzzy c-means(IFCM) clustering algorithm has a deficiency of easily falling into a local optimum, an improved IFCM which combines the traditional IFCM with the upgraded Genetic Algorithm(GA) is proposed. The traditional GA is upgraded from two aspects：the fitness standardization and the group diversification. The upgraded GA is more effective in global optimization, which can overcome the IFCM’s shortcoming of local optimum. Then the improved IFCM algorithm is innovatively practised in intrusion detection, and contrastive experiments on data sets KDD CUP99 show that, compared with IFCM algorithm, this algorithm advances the clustering precision effectively and has good reliability and feasibility.
Key words:
Intuitionistic Fuzzy c-means(IFCM),
clustering,
local optimum,
Genetic Algorithm(GA),
global optimization,
intrusion detection
中图分类号: