摘要
K-means算法在聚类分析中有着广泛应用。它采用了均值中心这一启发式信息,具有计算效率高的优点,但对初始聚类中心选择敏感,且容易陷入局部最优。PSO算法的随机性和并行性特点使其在处理数据库形式的海量数据中表现出更大的优越性,不仅具有较强的全局搜索能力,同时,通过对PSO算法搜索过程的改进增强了算法在最优解附近的搜索概率,降低样本对初始化敏感的程度,可以弥补K-means算法的缺陷。将改进的PSO算法应用于K-means聚类算法可以提高算法的稳定性和收敛效率,通过四组标准UCI数据集的试验,验证了新算法的有效性。
K- means algorithm,widely used in the clustering analysis,uses the mean center of the heuristic information and has the advantage of high computational efficiency. But it is sensitive to the initial center and easy to fall into local optimum. PSO algorithm,with the parallelism and randomness characteristics,has greater superiority in massive database processing. It not only has strong global searching capability,but also enhances the searching probability around the optimal solution. And it reduces sensitive level of the initialization,so PSO algorithm can compensate deficiencies for K- means.The improved PSO algorithm is applied to K- means clustering algorithm which improves the stability and convergence efficiency of the algorithm. The experiments on four standard UCI datasets demonstrate that the new algorithm has more effectiveness.
出处
《微处理机》
2016年第2期61-64,共4页
Microprocessors
基金
陕西省自然科学基金项目(No.2014JM8353)
关键词
K-平均算法
粒子群优化算法
聚类中心
稳定性
搜索
收敛
敏感
K-means
Particle Swarm Optimization
Clustering Center
Stability
Searching
Converging
Sensitive
