License
when quoting this document, please refer to the following
URN: urn:nbn:de:0030-drops-11332
URL: http://drops.dagstuhl.de/opus/volltexte/2007/1133/
Schneider, Petra

Relevance Matrices in LVQ

pdf-format:
Dokument 1.pdf (373 KB)

Abstract

LVQ-networks belong to the class of distance-based classifiers. The underlying distance measure is of special importance for their performance, because it defines how the data items are compared and how they are grouped in clusters. Relevance Learning techniques try to adapt the distance measure to the specific data used for training. I will present a new adaptive distance measure in Learning Vector Quantization which is an extension of previously proposed Relevance Learning schemes. In comparison to the already existing techniques for Relevance Learning, this distance measure is more powerful to represent the internal structure of the data appropriately. Two applications will be used to demonstrate the behavior of the new algorithm (artificial and real life).

BibTeX - Entry

@InProceedings{schneider:DSP:2007:1133,
  author =	{Petra Schneider},
  title =	{Relevance Matrices in LVQ},
  booktitle =	{Similarity-based Clustering and its Application to Medicine and Biology},
  year =	{2007},
  editor =	{Michael Biehl and Barbara Hammer and Michel Verleysen and Thomas Villmann },
  number =	{07131},
  series =	{Dagstuhl Seminar Proceedings},
  ISSN =	{1862-4405},
  publisher =	{Internationales Begegnungs- und Forschungszentrum f{\"u}r Informatik (IBFI), Schloss Dagstuhl, Germany},
  address =	{Dagstuhl, Germany},
  URL =		{http://drops.dagstuhl.de/opus/volltexte/2007/1133},
  annote =	{Keywords: Learning Vector Quantization, Relevance Learning, adaptive distance measure}
}

Keywords: Learning Vector Quantization, Relevance Learning, adaptive distance measure
Seminar: 07131 - Similarity-based Clustering and its Application to Medicine and Biology
Issue date: 2007
Date of publication: 12.09.2007

DROPS-Home | Fulltext Search | Imprint Published by LZI