CAT: Class Aware Adaptive Thresholding for Semi-Supervised Domain Generalization

TL;DR

This paper introduces CAT, a novel semi-supervised learning method for domain generalization that uses adaptive thresholding and label refinement to improve robustness and performance with limited labeled data across unseen domains.

Contribution

We propose CAT, a new approach combining adaptive thresholding and noisy label refinement for semi-supervised domain generalization, addressing limitations of fixed thresholds and label noise.

Findings

Outperforms existing methods on multiple benchmarks

Achieves high class diversity in pseudo-labels

Demonstrates robustness to noisy pseudo-labels

Abstract

Domain Generalization (DG) seeks to transfer knowledge from multiple source domains to unseen target domains, even in the presence of domain shifts. Achieving effective generalization typically requires a large and diverse set of labeled source data to learn robust representations that can generalize to new, unseen domains. However, obtaining such high-quality labeled data is often costly and labor-intensive, limiting the practical applicability of DG. To address this, we investigate a more practical and challenging problem: semi-supervised domain generalization (SSDG) under a label-efficient paradigm. In this paper, we propose a novel method, CAT, which leverages semi-supervised learning with limited labeled data to achieve competitive generalization performance under domain shifts. Our method addresses key limitations of previous approaches, such as reliance on fixed thresholds and sensitivity to noisy pseudo-labels. CAT combines adaptive thresholding with noisy label refinement techniques, creating a straightforward yet highly effective solution for SSDG tasks. Specifically, our approach uses flexible thresholding to generate high-quality pseudo-labels with higher class diversity while refining noisy pseudo-labels to improve their reliability. Extensive experiments across multiple benchmark datasets demonstrate the superior performance of our method, highlighting its effectiveness in achieving robust generalization under domain shift.