Learning vector quantization: cluster size and cluster number

Abstract

We study learning vector quantization methods to adapt the size of (hyper-)spherical clusters to better fit a given data set, especially in the context of non-normalized activations. The basic idea of our approach is to compute a desired radius from the data points that are assigned to a cluster in the direction of this desired radius. Since cluster size… (More)

Topics

4 Figures and Tables

Cite this paper

@article{Borgelt2004LearningVQ, title={Learning vector quantization: cluster size and cluster number}, author={Christian Borgelt and Daniela Girimonte and Giuseppe Acciani}, journal={2004 IEEE International Symposium on Circuits and Systems (IEEE Cat. No.04CH37512)}, year={2004}, volume={5}, pages={V-V} }