摘要
针对模糊C-均值算法(FCM)对初始值敏感的问题,提出禁忌搜索粒子群算法来优化FCM算法初始聚类中心.该混合算法是以粒子群算法为主体,禁忌算法针对粒子群算法的输出做更新,以避免单一使用粒子群算法而陷入局部最优的困境.算法保留了粒子群算法的并行处理能力,同时利用了禁忌搜索算法跳出局部最优解的特性,加快了整体算法的收敛速度并提高了聚类的准确率.
Aiming at the problem that the result is sensitive to the initial value selecting in FCM algorithm, an improved algorithm based on particle swarm optimization with tabu search was proposed. The hybrid al- gorithm combines PSO and TS algorithms. It makes use of TS algorithm updating the result of the PSO al- gorithm to avoid the algorithm trapping into the local optimum when using PSO algorithm alone. The hy- brid algorithm retains the parallel processing ability of the PSO algorithm, and has the character of the TS algorithm that can jump out the local optimum. The results of the experiment indicate that the hybrid algo- rithm speeds up the convergence of the whole algorithm and also improves the accuracy of the clustering.
出处
《湖北工业大学学报》
2013年第2期45-48,共4页
Journal of Hubei University of Technology
基金
湖北省科技厅国际合作项目(2010BFA007)
湖北省自然科学基金(2011CHB003)
关键词
模糊C-均值算法
粒子群算法
禁忌算法
fuzzy clustering algorithm
particle swarm optimization algorithm
tabu search algorithm
