摘要
传统的k-均值算法对初始聚类中心的敏感很大,极易陷入局部最优值;利用遗传算法或免疫规划算法解决初始聚类中心是较好的方法,但后期容易出现收敛速度缓慢。为了克服上述缺点,文章将免疫原理的选择操作机制引入遗传算法中,使个体浓度和适应度同时对个体的选择施加影响,以此提出基于改进遗传算法的K-均值聚类算法,该方法利用K-均值算法的高效性和改进遗传算法的全局优化搜索能力,较好地解决了聚类中心优化问题。试验结果表明,本算法能够有效改善聚类质量,并且具有较好的收敛速度。
The traditional K - means algorithm has the shortcoming that plunges into a local optimum prematurely because of sensitive selection of the initial cluster center. Using the genetic or immune algorithm into K - means algorithm to optimize cluster center is much better than using other algorithms, but there appeares the local early phenomenon easily. In order to overcome the shortcomings men- tioned above, a K -means clustering algorithm based on improved Genetic Algorithm is proposed, which useing the advantages of im- mune idea and introducing the idea of selection opreation of immune principle into Genetic Algorithm,in which the selection of individu- al was impacted by its density and fitness. The algorithm can solve the problem of optimizing cluster center by combining the high effi- ciency of K - means algorithm with the ability of global optimization of impoved Genetic Algorithm. The experimental results show that new algorithm has improved the clustering quality effectively, and greater global searching capability.
出处
《微计算机应用》
2010年第4期11-15,共5页
Microcomputer Applications
基金
河南省科技攻关计划项目(082102210064)
河南省教育厅科研项目资助(2008A510007)
关键词
聚类分析
遗传算法
免疫机制
K-均值
个体浓度
clustering analysis, genetic algorithm, immune principle, K - means algorithm, individual density
