Skip to Main content Skip to Navigation
Conference papers

PAC-Bayes Bounds for the Risk of the Majority Vote

Abstract : We propose new PAC-Bayes bounds for the risk of the weighted majority vote that depend on the mean and variance of the error of its associated Gibbs classifier. We show that these bounds can be smaller than the risk of the Gibbs classifier and can be arbitrarily close to zero even if the risk of the Gibbs classifier is close to 1/2. Moreover, we show that these bounds can be uniformly estimated on the training data for all possible posteriors Q. Moreover, they can be improved by using a large sample of unlabelled data.
Document type :
Conference papers
Complete list of metadata

https://hal.archives-ouvertes.fr/hal-01352012
Contributor : Lip6 Publications <>
Submitted on : Friday, August 5, 2016 - 11:10:48 AM
Last modification on : Friday, January 8, 2021 - 5:34:12 PM

Identifiers

  • HAL Id : hal-01352012, version 1

Citation

Alexandre Lacasse, François Laviolette, Mario Marchand, Pascal Germain, Nicolas Usunier. PAC-Bayes Bounds for the Risk of the Majority Vote. Advances in Neural Information Processing Systems (NIPS'06), Dec 2006, Vancouver, Canada. pp.769-776. ⟨hal-01352012⟩

Share

Metrics

Record views

72