Causality-based Dual-Contrastive Learning Framework for Domain Generalization

TL;DR

This paper introduces a causality-based dual-contrastive learning framework for domain generalization, combining feature and prototype contrast, a causal fusion attention module, and a hard-pair mining strategy to improve generalization to unseen domains.

Contribution

It proposes a novel dual-contrastive learning approach with a causal fusion attention module and hard-pair mining, enhancing domain generalization without complex architectures.

Findings

Outperforms state-of-the-art on three DG datasets.

Effective without domain labels as a plug-and-play module.

Improves robustness against domain shifts.

Abstract

Domain Generalization (DG) is essentially a sub-branch of out-of-distribution generalization, which trains models from multiple source domains and generalizes to unseen target domains. Recently, some domain generalization algorithms have emerged, but most of them were designed with non-transferable complex architecture. Additionally, contrastive learning has become a promising solution for simplicity and efficiency in DG. However, existing contrastive learning neglected domain shifts that caused severe model confusions. In this paper, we propose a Dual-Contrastive Learning (DCL) module on feature and prototype contrast. Moreover, we design a novel Causal Fusion Attention (CFA) module to fuse diverse views of a single image to attain prototype. Furthermore, we introduce a Similarity-based Hard-pair Mining (SHM) strategy to leverage information on diversity shift. Extensive experiments show that our method outperforms state-of-the-art algorithms on three DG datasets. The proposed algorithm can also serve as a plug-and-play module without usage of domain labels.