Clustering and Aggregation of Relational Data with Applications to Image Database Categorization

H. Frigui and C. Hwang (USA)


Relational Clustering; Feature aggregation; Image database categorization.


In this paper, we introduce a new algorithm for Clustering and Aggregating Relational Data (CARD). We assume that data is available in a relational form, where we only have information about the degrees to which pairs of objects in the data set are related. Moreover, we assume that the rela tional information is represented by multiple dissimilarity matrices. These matrices could have been generated us ing different sensors, features, or mappings. CARD is de signed to aggregate pairwise distances from multiple rela tional matrices, partition the data into clusters, and learn a relevance weight for each matrix in each cluster simultane ously. The cluster dependent relevance weights offer two advantages. First, they guide the clustering process to par tition the data set into more meaningful clusters. Second, they can be used in subsequent steps of a learning system to improve its learning behavior. The performance of the proposed algorithm is illustrated by using it to categorize a collection of 4000 color images. We represent the pairwise image dissimilarities by four different relational matrices that encode the color, texture, and structure information.

Important Links:

Go Back