Fuzzy Clustering with Pairwise Constraints for Knowledge-Driven Image Categorization

The identification of categories in image databases usually relies on clustering algorithms that only exploit the feature-based similarities between images. The addition of semantic information should help improving the results of the categorization process. Pairwise constraints between some images are easy to provide, even when the user has a very incomplete prior knowledge of the image categories he/she can expect to find in a database. A categorization approach relying on such semantic information is called semi-supervised clustering. We present here a new semi-supervised clustering algorithm, Pairwise-Constrained Competitive Agglomeration, based on a fuzzy cost function that takes pairwise constraints into account. Our evaluations show that with a rather low number of constraints this algorithm can significantly improve the categorization.

• Type: Conference
• Year: 2004
• Where: EWIMT
• Authors: Nizar Grira, Michel Crucianu, Nozha Boujemaa
• Keywords: Clustering Algorithm, EWIMT 2004, Knowledge Management, Semantic Information, Semi-supervised Clustering