DUAL OBJECT-RELATION CLUSTERING MODELS
James C. Bezdek, Richard J. Hathaway · International Journal of General Systems · 1990
In this paper we review some fuzzy clustering algorithms that are defined by optimization of penalty functions. The discussion is divided into two parts: models for and algorithms that process numerical object (individual) data; and those that are designed for (pairwise) numerical relational data. We discuss some problems and benefits associated with finding dual models and algorithms for the two kinds of data, and exemplify this idea with the object and relational c-means clustering duals.