DMCA
Selecting Input Variables Using Mutual Information and Nonparametric Density Estimation (1996)
| Citations: | 61 - 2 self |
BibTeX
@INPROCEEDINGS{Bonnlander96selectinginput,
author = {Brian V. Bonnlander and Andreas S. Weigend},
title = {Selecting Input Variables Using Mutual Information and Nonparametric Density Estimation},
booktitle = {},
year = {1996},
pages = {42--50},
publisher = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In learning problems where a connectionist network is trained with a finite sized training set, better generalization performance is often obtained when unneeded weights in the network are eliminated. One source of unneeded weights comes from the inclusion of input variables that provide little information about the output variables. We propose a method for identifying and eliminating these input variables. The method first determines the relationship between input and output variables using nonparametric density estimation and then measures the relevance of input variables using the information theoretic concept of mutual information. We present results from our method on a simple toy problem and a nonlinear time series.
Keyphrases
nonparametric density estimation input variable input variable using mutual information unneeded weight output variable simple toy problem mutual information nonlinear time series generalization performance introduction generalization performance information theoretic concept connectionist network poor generalization performance finite sized training set baum haussler little information many free parameter free parameter fixed-size training set present result
