Multi-step domain adaptation by adversarial attack to $\mathcal{H} \Delta \mathcal{H}$-divergence

TL;DR

This paper introduces a novel multi-step domain adaptation method leveraging adversarial examples' transferability to improve classifier accuracy across domains, demonstrating effectiveness on benchmark datasets.

Contribution

It proposes using adversarial attacks to reduce divergence between source and target domains, enhancing unsupervised domain adaptation performance.

Findings

Improved accuracy on Digits dataset

Enhanced performance on Office-Home dataset

Method integrates with existing domain adaptation techniques

Abstract

Adversarial examples are transferable between different models. In our paper, we propose to use this property for multi-step domain adaptation. In unsupervised domain adaptation settings, we demonstrate that replacing the source domain with adversarial examples to $\mathcal{H} \Delta \mathcal{H}$-divergence can improve source classifier accuracy on the target domain. Our method can be connected to most domain adaptation techniques. We conducted a range of experiments and achieved improvement in accuracy on Digits and Office-Home datasets.