109
click to vote
ICML
2001
IEEE
16 years 3 months ago
2001
IEEE
We present an improved bound on the difference between training and test errors for voting classifiers. This improved averaging bound provides a theoretical justification for popu...