摘要
聚类是发现数据分布和隐含模式的一项重要技术,但单一聚类算法却很难达到预期的效果。在缺乏样本集先验知识的前提下,目前的分类融合技术很难应用到聚类技术中,导致聚类融合技术起步很晚。近几年的研究发现,聚类融合方法对提高聚类算法的稳定性和高效性发挥了重要的作用。文中对近年来聚类融合的方法和国内外研究现状进行了简单综述,并且以基于投票的聚类融合算法为例,实验证明了其比单一聚类算法的优越性,展望了聚类融合算法的未来。
Clustering is a technique for the discovery of data distribution and latent data pattern.Single clustering is hard to reach good result.However,in unsupervised learning,researches of ensemble approaches are concerned only in recent years.Because of the premise of the prior knowledge of sample sets is insensible,the ensemble approaches of classifier can＇t be utilized in the same way directly. Recent studies proof that clustering ensemble approaches can enhance robustness and stabilities greatly.Makes an overview and the research status at home and abroad of the clustering ensemble approaches.It makes an example of clustering ensemble algorithm based on voting,show superiority than single.Finally,prospect the future of clustering ensemble approaches.
出处
《计算机技术与发展》
2010年第7期106-108,113,共4页
Computer Technology and Development
基金
安徽省自然科学重点资助项目(KJ2007A051)
关键词
聚类
融合技术
差异度
投票
clustering
ensemble technique
diversity
voting
