DomCLP: Domain-wise Contrastive Learning with Prototype Mixup for Unsupervised Domain Generalization

TL;DR

This paper introduces DomCLP, a novel unsupervised domain generalization method combining domain-wise contrastive learning and prototype mixup to improve the extraction of domain-irrelevant features, outperforming existing approaches.

Contribution

The paper proposes DomCLP, a new approach that enhances domain-irrelevant feature learning without strong assumptions, improving generalization across unseen domains.

Findings

Outperforms state-of-the-art on PACS and DomainNet datasets

Significant improvements across various label fractions

Effectively extracts domain-irrelevant features

Abstract

Self-supervised learning (SSL) methods based on the instance discrimination tasks with InfoNCE have achieved remarkable success. Despite their success, SSL models often struggle to generate effective representations for unseen-domain data. To address this issue, research on unsupervised domain generalization (UDG), which aims to develop SSL models that can generate domain-irrelevant features, has been conducted. Most UDG approaches utilize contrastive learning with InfoNCE to generate representations, and perform feature alignment based on strong assumptions to generalize domain-irrelevant common features from multi-source domains. However, existing methods that rely on instance discrimination tasks are not effective at extracting domain-irrelevant common features. This leads to the suppression of domain-irrelevant common features and the amplification of domain-relevant features, thereby hindering domain generalization. Furthermore, strong assumptions underlying feature alignment can lead to biased feature learning, reducing the diversity of common features. In this paper, we propose a novel approach, DomCLP, Domain-wise Contrastive Learning with Prototype Mixup. We explore how InfoNCE suppresses domain-irrelevant common features and amplifies domain-relevant features. Based on this analysis, we propose Domain-wise Contrastive Learning (DCon) to enhance domain-irrelevant common features. We also propose Prototype Mixup Learning (PMix) to generalize domain-irrelevant common features across multiple domains without relying on strong assumptions. The proposed method consistently outperforms state-of-the-art methods on the PACS and DomainNet datasets across various label fractions, showing significant improvements. Our code will be released. Our project page is available at https://github.com/jinsuby/DomCLP.