Multi-View Fuzzy Clustering with Minimax Optimization for Effective Clustering of Data from Multiple Sources

摘要

Multi-view data clustering refers to categorizing a data set by making gooduse of related information from multiple representations of the data. Itbecomes important nowadays because more and more data can be collected in avariety of ways, in different settings and from different sources, so each dataset can be represented by different sets of features to form different views ofit. Many approaches have been proposed to improve clustering performance byexploring and integrating heterogeneous information underlying different views.In this paper, we propose a new multi-view fuzzy clustering approach calledMinimaxFCM by using minimax optimization based on well-known Fuzzy c means. InMinimaxFCM the consensus clustering results are generated based on minimaxoptimization in which the maximum disagreements of different weighted views areminimized. Moreover, the weight of each view can be learned automatically inthe clustering process. In addition, there is only one parameter to be setbesides the fuzzifier. The detailed problem formulation, updating rulesderivation, and the in-depth analysis of the proposed MinimaxFCM are providedhere. Experimental studies on nine multi-view data sets including real worldimage and document data sets have been conducted. We observed that MinimaxFCMoutperforms related multi-view clustering approaches in terms of clusteringaccuracy, demonstrating the great potential of MinimaxFCM for multi-view dataanalysis.