Free Lunch for Domain Adversarial Training: Environment Label Smoothing

TL;DR

This paper introduces Environment Label Smoothing (ELS), a technique that stabilizes domain adversarial training by softening discriminator outputs, thereby improving robustness and performance in domain generalization tasks with noisy environment labels.

Contribution

The paper proposes Environment Label Smoothing (ELS) to enhance domain adversarial training stability and robustness against noisy environment labels, supported by theoretical analysis and experimental validation.

Findings

ELS improves training stability and convergence.

ELS enhances robustness to noisy environment labels.

State-of-the-art results achieved on domain generalization tasks.

Abstract

A fundamental challenge for machine learning models is how to generalize learned models for out-of-distribution (OOD) data. Among various approaches, exploiting invariant features by Domain Adversarial Training (DAT) received widespread attention. Despite its success, we observe training instability from DAT, mostly due to over-confident domain discriminator and environment label noise. To address this issue, we proposed Environment Label Smoothing (ELS), which encourages the discriminator to output soft probability, which thus reduces the confidence of the discriminator and alleviates the impact of noisy environment labels. We demonstrate, both experimentally and theoretically, that ELS can improve training stability, local convergence, and robustness to noisy environment labels. By incorporating ELS with DAT methods, we are able to yield state-of-art results on a wide range of domain generalization/adaptation tasks, particularly when the environment labels are highly noisy.