Continual Domain Adversarial Adaptation via Double-Head Discriminators

TL;DR

This paper introduces a double-head discriminator method for continual domain adversarial adaptation, effectively reducing estimation errors and improving target domain performance while mitigating source domain forgetting.

Contribution

The paper proposes a novel double-head discriminator approach with a source-only discriminator to enhance continual domain adaptation under limited source data.

Findings

Achieves over 2% improvement across target categories.

Reduces empirical $ ext{H}$-divergence estimation error.

Mitigates forgetting on source domain.

Abstract

Domain adversarial adaptation in a continual setting poses a significant challenge due to the limitations on accessing previous source domain data. Despite extensive research in continual learning, the task of adversarial adaptation cannot be effectively accomplished using only a small number of stored source domain data, which is a standard setting in memory replay approaches. This limitation arises from the erroneous empirical estimation of $\gH$-divergence with few source domain samples. To tackle this problem, we propose a double-head discriminator algorithm, by introducing an addition source-only domain discriminator that are trained solely on source learning phase. We prove that with the introduction of a pre-trained source-only domain discriminator, the empirical estimation error of $\gH$-divergence related adversarial loss is reduced from the source domain side. Further experiments on existing domain adaptation benchmark show that our proposed algorithm achieves more than 2$\%$ improvement on all categories of target domain adaptation task while significantly mitigating the forgetting on source domain.