Adaptation of Cluster Sizes in Objective Function Based Fuzzy Clustering Technology

Kurzfassung

This paper discusses new approaches in objective function based fuzzy clustering. Some well-known approaches are extended by a supplementary component. The resulting new clustering techniques are able to adapt single clusters to the expansion of the corresponding group of data in an iterative optimization procedure. Another new approach based on volume centers as cluster representatives with varying radii for individual groups is also described. The corresponding objective functions are presented and alternating optimization schemes are derived. Experimental results demonstrate the significance of the presented techniques.