DiCoTTA: Domain-invariant Learning for Continual Test-time Adaptation

2025-04-07

AI Summary

This work addresses the challenge in continual test-time adaptation (CTTA) where models struggle to simultaneously achieve online adaptation to the current test domain and generalization to future unseen domains. We propose the first online domain-invariant learning framework for CTTA. Our method jointly disentangles feature representations across current and historical test domains, preserving semantic integrity while enabling knowledge retention. It incorporates a lightweight model architecture, a domain-invariant test-time optimization strategy, an incremental domain memory module, and a dual-objective online optimization algorithm. Evaluated on four standard CTTA benchmarks, our approach achieves state-of-the-art performance, significantly improving generalization to unseen test domains, robustness to distribution shifts, and adaptation stability—overcoming the fundamental limitation of conventional CTTA methods that adapt exclusively to the current domain.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising

Abstract

This paper studies continual test-time adaptation (CTTA), the task of adapting a model to constantly changing unseen domains in testing while preserving previously learned knowledge. Existing CTTA methods mostly focus on adaptation to the current test domain only, overlooking generalization to arbitrary test domains a model may face in the future. To tackle this limitation, we present a novel online domain-invariant learning framework for CTTA, dubbed DiCoTTA. DiCoTTA aims to learn feature representation to be invariant to both current and previous test domains on the fly during testing. To this end, we propose a new model architecture and a test-time adaptation strategy dedicated to learning domain-invariant features without corrupting semantic contents, along with a new data structure and optimization algorithm for effectively managing information from previous test domains. DiCoTTA achieved state-of-the-art performance on four public CTTA benchmarks. Moreover, it showed superior generalization to unseen test domains.

Problem

• Adapting models to constantly changing unseen test domains
• Preserving learned knowledge while adapting to new domains
• Ensuring generalization to arbitrary future test domains

Innovation

• Online domain-invariant learning framework
• Model architecture for invariant features
• Optimization algorithm managing previous domains