选择性聚类融合新方法研究

New algorithm for selective clustering ensemble

针对传统选择性聚类融合算法不能消除劣质聚类成员的干扰以及聚类准确性不高等问题,提出了一种新的选择性加权聚类融合算法。算法中提出了基于聚类有效性评价方法的参照成员选择方法和联合聚类质量以及差异度的选择策略,然后还提出了基于容错关系信息熵的属性重要性加权方法。新算法有效地克服了传统选择性聚类融合算法的缺点,消除了劣质聚类成员的干扰,提高了聚类的准确性。大量的对比实验结果表明了算法的有效,且性能显著提高。

Traditional selective clustering ensemble doesn＇t eliminate the inferior quality＇influence and the accuracy of clustering is not high.In order to solve these problem,this paper proposed a new selective clustering ensemble algorithm.The algorithm,used clustering validity evaluation to evaluate all available clustering ensemble partitions and selected the best quality as reference partition.Secondly,it defined selection strategy via the quality and diversity.Lastly,this paper proposed setting weights to ensemble members according to the significance of attribute in tolerance relation theory.The experimental results show that the new algorithm is effective and clustering performance can be significantly improved.