摘要
1.引言
聚类分析(clustering)是人工智能研究的重要领域.聚类方法被广泛研究并应用于机器学习、统计分析、模式识别以及数据库数据挖掘与知识发现等不同的领域.
As one of the most popular clustering techniques, K-Means algorithm usually obtains locally optimal solutions due to its sensitivity to initial starting center. To overcome this problem, a genetic algorithm is used to search the initial center for K-Means algorithm. A concept of 'Gene difference' is introduced to control the crossover operator and mutation operator in genetic algorithm. Experiments on standard database of UCI show that the proposed method can efficiently improve the clustering result.
出处
《计算机科学》
CSCD
北大核心
2002年第7期94-96,共3页
Computer Science
基金
天津市自然科学基金(003600311)
