Algorithms and Theory for Multiple-Source Adaptation

TL;DR

This paper introduces novel algorithms and theoretical guarantees for multiple-source adaptation, demonstrating improved robustness and performance across diverse target distributions through comprehensive experiments.

Contribution

It provides new normalized solutions with strong theoretical guarantees and algorithms for distribution-weighted combinations in multiple-source adaptation.

Findings

Our algorithm outperforms competing methods on real-world datasets.

Theoretical guarantees hold even with distinct source domain conditional probabilities.

A single robust model performs well across various target mixtures.

Abstract

This work includes a number of novel contributions for the multiple-source adaptation problem. We present new normalized solutions with strong theoretical guarantees for the cross-entropy loss and other similar losses. We also provide new guarantees that hold in the case where the conditional probabilities for the source domains are distinct. Moreover, we give new algorithms for determining the distribution-weighted combination solution for the cross-entropy loss and other losses. We report the results of a series of experiments with real-world datasets. We find that our algorithm outperforms competing approaches by producing a single robust model that performs well on any target mixture distribution. Altogether, our theory, algorithms, and empirical results provide a full solution for the multiple-source adaptation problem with very practical benefits.