Active Semi-supervised Clustering for Image Database Categorization

Clustering algorithms are increasingly employed for the categorization of image databases, in order to provide users with database overviews and make their access more effective. By including information provided by the user, the categorization process can produce results that come closer to user’s expectations. To make such a semi-supervised categorization approach acceptable for the user, this information must be of a very simple nature and the amount of information the user is required to provide should be minimized. For a semi-supervised fuzzy clustering algorithm we developed, Pairwise-Constrained Competitive Agglomeration, we put forward here a criterion for the active selection of constraints. We show that this selection criterion allows a significant reduction in the number of pairwise constraints required, making the resulting algorithm an attractive alternative in the categorization of image databases.