Concept-Based Unsupervised Domain Adaptation

TL;DR

This paper introduces CUDA, a framework that enhances concept bottleneck models for domain adaptation by aligning concept representations across domains, allowing for minor differences, and inferring concepts in unlabeled target domains, thus improving robustness and interpretability.

Contribution

The paper proposes CUDA, a novel method combining adversarial training and relaxed constraints to enable unsupervised domain adaptation for concept bottleneck models with theoretical guarantees.

Findings

CUDA outperforms existing methods on real-world datasets.

It effectively aligns concept representations across domains.

The approach maintains interpretability while improving robustness.

Abstract

Concept Bottleneck Models (CBMs) enhance interpretability by explaining predictions through human-understandable concepts but typically assume that training and test data share the same distribution. This assumption often fails under domain shifts, leading to degraded performance and poor generalization. To address these limitations and improve the robustness of CBMs, we propose the Concept-based Unsupervised Domain Adaptation (CUDA) framework. CUDA is designed to: (1) align concept representations across domains using adversarial training, (2) introduce a relaxation threshold to allow minor domain-specific differences in concept distributions, thereby preventing performance drop due to over-constraints of these distributions, (3) infer concepts directly in the target domain without requiring labeled concept data, enabling CBMs to adapt to diverse domains, and (4) integrate concept learning into conventional domain adaptation (DA) with theoretical guarantees, improving interpretability and establishing new benchmarks for DA. Experiments demonstrate that our approach significantly outperforms the state-of-the-art CBM and DA methods on real-world datasets.