Abstract
An algorithm is presented that predicts the mean recognition accuracy as a function of dimensionality for two class problems, using a Bayes classifier in the presence of a limited number of training samples. Several experiments are presented to assess the algorithm's performance, and a binary tree classification procedure that utilizes the algorithm is shown to prove its usefulness.
LARS Tech Report Number
070882
Date of this Version
January 1982
Recommended Citation
Muasher, M. J. and Landgrebe, D. A., "A Binary Tree Feature Selection Technique for Limited Training Sample Size" (1982). LARS Technical Reports. Paper 84.
https://docs.lib.purdue.edu/larstech/84
COinS