Multiple Source Adaptation and the Renyi Divergence

TL;DR

This paper provides a comprehensive theoretical analysis of multiple source adaptation using Renyi divergence, extending previous bounds to arbitrary target distributions and approximate source distributions, with experimental validation.

Contribution

It broadens the scope of multiple source adaptation analysis by considering arbitrary target distributions, approximate sources, and different labeling functions, with new theoretical bounds and experimental results.

Findings

Distribution weighted combinations improve adaptation performance.

Renyi divergence-based bounds are effective for both known and unknown target distributions.

Experimental results validate the theoretical bounds and show performance benefits.

Abstract

This paper presents a novel theoretical study of the general problem of multiple source adaptation using the notion of Renyi divergence. Our results build on our previous work [12], but significantly broaden the scope of that work in several directions. We extend previous multiple source loss guarantees based on distribution weighted combinations to arbitrary target distributions P, not necessarily mixtures of the source distributions, analyze both known and unknown target distribution cases, and prove a lower bound. We further extend our bounds to deal with the case where the learner receives an approximate distribution for each source instead of the exact one, and show that similar loss guarantees can be achieved depending on the divergence between the approximate and true distributions. We also analyze the case where the labeling functions of the source domains are somewhat different. Finally, we report the results of experiments with both an artificial data set and a sentiment analysis task, showing the performance benefits of the distribution weighted combinations and the quality of our bounds based on the Renyi divergence.