| | |
Summary: An Informational Energy LVQ Approach for
Feature Ranking
Razvan Andonie 1
and Angel Cat¸aron 2
1
Computer Science Department, Central Washington University, USA
2
Electronics and Computers Department, Transylvania University, Romania
Abstract. Input feature ranking and selection represent a necessary
preprocessing stage in classification, especially when one is required to
manage large quantities of data. We introduce a weighted LVQ algo-
rithm, called Energy Relevance LVQ (ERLVQ), based on Onicescu's in-
formational energy [10]. ERLVQ is an incremental learning algorithm for
supervised classification and feature ranking.
1 Introduction
Standard Learning Vector Quantization (LVQ) [9] does not discriminate be-
tween more or less informative features: their influence to the distance function
is equal. On the contrary, the Distinction Sensitive Learning Vector Quantizer
(DSLVQ), introduced by Pregenzer et al. [11], holds a changeable weight (rel-
evance) value for every feature and employs a weighted distance function for
|
