A New PAC-Bayesian Perspective on Domain Adaptation
This work addresses domain adaptation for machine learning practitioners by providing a theoretical framework and algorithm, though it appears incremental as it builds on existing PAC-Bayesian methods.
The paper tackles domain adaptation by deriving a PAC-Bayesian upper-bound on target risk that uses a distribution divergence ratio to balance source error and target disagreement, leading to a learning algorithm tested on real data.
We study the issue of PAC-Bayesian domain adaptation: We want to learn, from a source domain, a majority vote model dedicated to a target one. Our theoretical contribution brings a new perspective by deriving an upper-bound on the target risk where the distributions' divergence---expressed as a ratio---controls the trade-off between a source error measure and the target voters' disagreement. Our bound suggests that one has to focus on regions where the source data is informative.From this result, we derive a PAC-Bayesian generalization bound, and specialize it to linear classifiers. Then, we infer a learning algorithmand perform experiments on real data.