Attention-based Cross-Layer Domain Alignment for Unsupervised Domain Adaptation

TL;DR

This paper introduces ACDA, an attention-based method that aligns semantic features across different model layers to improve unsupervised domain adaptation, achieving state-of-the-art results.

Contribution

The paper proposes a novel attention mechanism for cross-layer semantic alignment, enhancing domain adaptation performance over existing methods.

Findings

ACDA outperforms existing methods on multiple benchmarks.

The attention mechanism effectively reweights cross-layer semantic similarities.

ACDA achieves state-of-the-art accuracy in unsupervised domain adaptation.

Abstract

Unsupervised domain adaptation (UDA) aims to learn transferable knowledge from a labeled source domain and adapts a trained model to an unlabeled target domain. To bridge the gap between source and target domains, one prevailing strategy is to minimize the distribution discrepancy by aligning their semantic features extracted by deep models. The existing alignment-based methods mainly focus on reducing domain divergence in the same model layer. However, the same level of semantic information could distribute across model layers due to the domain shifts. To further boost model adaptation performance, we propose a novel method called Attention-based Cross-layer Domain Alignment (ACDA), which captures the semantic relationship between the source and target domains across model layers and calibrates each level of semantic information automatically through a dynamic attention mechanism. An elaborate attention mechanism is designed to reweight each cross-layer pair based on their semantic similarity for precise domain alignment, effectively matching each level of semantic information during model adaptation. Extensive experiments on multiple benchmark datasets consistently show that the proposed method ACDA yields state-of-the-art performance.