preprint and report available

[Michael Biehl]

Dear colleagues,
 the following preprint and a related technical report are
 now available on-line at

 http://www.cs.rug.nl/~biehl/prepneuro.html

 The dynamics of Learning Vector Quantization
   M. Biehl, A. Ghosh, and B. Hammer
   accepted contribution to the European Symposium on
   Artificial Neural Networks, Bruges, 2005

 A theoretical framework for analysing the dynamics of LVQ
   M. Biehl, A. Freking, A. Ghosh, and G. Reents
   Technical Rep. 2004-9-02, Mathematics and Computing Science
   Univ. Groningen, P.O. 800, 9700 AV Groningen, The Netherlands


 Abstract of the preprint:

 Winnter Takes All (WTA) algorithms offer intuitive and powerful
 learning schemes such as Learning Vector Quantization (LVQ) and
 variations thereof, most of which are heuristically motivated.
 In this article we investigate in an exact mathematical way the
 dynamics of different Vector Quantization (VQ) schemes including
 standard LVQ. We consider the training from high-dimensional data
 generated according to a mixture of overlapping Gaussians and the
 case of two prototypes. Simplifying assumptions allow for an exact
 description of the on-line learning dynamics of the learning
 processes and the achievable generalization error.



------------------------------------------------------

    Michael Biehl

    Rijksuniversiteit Groningen
    Wiskunde & Informatica
    Blauwborgje 3,  9747 AC Groningen
    The Netherlands

    e-mail biehl at cs.rug.nl
    web    www.cs.rug.nl/~biehl