Evaluating the Effectiveness of Soft K-Means in Detecting Overlapping Clusters.
ICTCS(2016)
摘要
Traditional clustering techniques are limited in detecting clusters of overlapping in nature. To identify overlapping clusters in a real data is a challenging task in data mining. A number of clustering methods proposed so far to address the issue. They have their own merits and limitations in producing quality outcomes. Soft computing is relatively a new computing paradigm to handle uncertainty and vagueness in the data. Recently, soft computing concepts are incorporated in data clustering where data are overlapping in distribution. Various literatures reveal that soft clustering techniques are effective and give promising results. However, very few soft clustering approaches are evolved so far. Most of them are the modified and hybrid version of well-known k-means algorithm. In this paper, we evaluate the performance of all soft versions of k-means such as fuzzy, rough and fuzzy-rough c-means in the light of synthetic data to measure its effectiveness in detecting overlapping clusters.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络