NEURAL CLUSTERING NETWORKS BASED ON GLOBAL OPTIMIZATION OF PROTOTYPESIN METRIC-SPACES

The utilisation of clustering algorithms based on the optimisation of prototypes in neural networks is demonstrated for unsupervised learnin g. Stimulated by common clustering methods of this type (learning vect or quantisation [LVQ, GLVQ] and K-means) a globally operating algorith m was developed to cope with known shortcomings of existing tools. Thi s algorithm and K-means (for the common methods) were applied to the p roblem of clustering EEG patterns being preprocessed. It can be shown that the algorithm based on global random optimisation may find an opt imal solution repeatedly, whereas K-means provides different sub-optim al solutions with respect to the quality measure defined as objective function. The results are presented. The performance of the algorithms is discussed.