Event Selection Using an Extended Fisher Discriminant Method
Data Analysis, Statistics and Probability
2007-05-23 v1
Abstract
This note discusses the problem of choosing between hypotheses in a situation with many, correlated non-normal variables. A new method is introduced to shrink the many variables into a smaller subset of variables with zero mean, unit variance, and zero correlation coefficient between variables. These new variables are well suited to use in a neural net.
Cite
@article{arxiv.physics/0310145,
title = {Event Selection Using an Extended Fisher Discriminant Method},
author = {Byron P. Roe},
journal= {arXiv preprint arXiv:physics/0310145},
year = {2007}
}
Comments
Paper submitted to Stanford PHYSTAT 2003 conference, 3pages 2 figures