Detecting Ineffective Features for Pattern Recognition

View/ Open
Date
2017-10-17MFO Scientific Program
Research in Pairs 2017Series
Oberwolfach Preprints;2017,26Author
Györfi, László
Walk, Harro
Metadata
Show full item recordOWP-2017-26
Abstract
For a binary classification problem, the hypothesis testing is studied, that a component of the observation vector is not effective, i.e., that component carries no information for the classification. We introduce nearest neighbor and partitioning estimates of the Bayes error probability, which result in a strongly consistent test.