Rethinking Distributional Matching Based Domain Adaptation

TL;DR

This paper critically examines the limitations of distributional matching in domain adaptation under realistic shifts and introduces InstaPBM, a new instance-based method that significantly improves robustness and accuracy in diverse benchmarks.

Contribution

The paper identifies the failure of existing distributional matching methods under realistic domain shifts and proposes InstaPBM, a novel instance-based approach that enhances robustness in domain adaptation.

Findings

InstaPBM outperforms baselines by 2.2-4.5% in accuracy across multiple benchmarks.

Distributional matching methods often fail under realistic domain shifts.

InstaPBM demonstrates significant improvements in robustness and accuracy in both conventional and realistic domain shift scenarios.

Abstract

Domain adaptation (DA) is a technique that transfers predictive models trained on a labeled source domain to an unlabeled target domain, with the core difficulty of resolving distributional shift between domains. Currently, most popular DA algorithms are based on distributional matching (DM). However in practice, realistic domain shifts (RDS) may violate their basic assumptions and as a result these methods will fail. In this paper, in order to devise robust DA algorithms, we first systematically analyze the limitations of DM based methods, and then build new benchmarks with more realistic domain shifts to evaluate the well-accepted DM methods. We further propose InstaPBM, a novel Instance-based Predictive Behavior Matching method for robust DA. Extensive experiments on both conventional and RDS benchmarks demonstrate both the limitations of DM methods and the efficacy of InstaPBM: Compared with the best baselines, InstaPBM improves the classification accuracy respectively by $4.5\%$, $3.9\%$ on Digits5, VisDA2017, and $2.2\%$, $2.9\%$, $3.6\%$ on DomainNet-LDS, DomainNet-ILDS, ID-TwO. We hope our intuitive yet effective method will serve as a useful new direction and increase the robustness of DA in real scenarios. Code will be available at anonymous link: https://github.com/pikachusocute/InstaPBM-RobustDA.