6 - PAC-Bayesian Theory for Domain Adaptation

Abstract

Listen

Translate article

As recalled in section 1.2.3, the PAC–Bayesian theory offers tools to derive generalization bounds for models that take the form of a weighted majority vote over a set of functions that can be the hypothesis space. In this section, we recall the work done by Germain et al. on how the PAC–Bayesian theory can help to theoretically understand domain adaptation through the weighted majority vote learning point of view.