A new cluster isolation criterion based on dissimilarity increments

Abstract

Abstract-This paper addresses the problem of cluster defining criteria by proposing a model-based characterization of interpattern relationships. Taking a dissimilarity matrix between patterns as the basic measure for extracting group structure, dissimilarity increments between neighboring patterns within a cluster are analyzed. Empirical evidence suggests modeling the statistical distribution of these increments by an exponential density; we propose to use this statistical model, which characterizes context, to derive a new cluster isolation criterion. The integration of this criterion in a hierarchical agglomerative clustering framework produces a partitioning of the data, while exhibiting data interrelationships in terms of a dendrogram-type graph. The analysis of the criterion is undertaken through a set of examples, showing the versatility of the method in identifying clusters with arbitrary shape and size; the number of clusters is intrinsically found without requiring ad hoc specification of design parameters nor engaging in a computationally demanding optimization procedure.

A new validity index for crisp clusters

“…Then, cluster validity indices are used to find the best partitioning of data. A great number of such indices have been introduced, e.g., [3,9,10,13,17,26,28,30,31]. In many validity indices two properties of clusters are taken into account, i.e., compactness and separability [11].…”

A density-based cluster validity approach using multi-representatives

“…We also introduce the idea of evaluating the density changes within clusters in order to estimate the clusters' cohesion. The cohesion concept used in this work seems very close to the path-based clustering criteria proposed by Fred and Leitao (2003). However these criteria are based on distances and measure the dissimilarity increments between clusters contrary to our cohesion criterion that estimates density changes within clusters.…”

Improving Image Classification Quality Via Dissimilarity Measure In Non-Euclidean Spaces

“…According to the introduction in literature [10], the dissimilarity diffusion between these patterns is formulated using the following equation:…”