CTTA-T: Continual Test-Time Adaptation for Text Understanding via Teacher-Student with a Domain-aware and Generalized Teacher

2025-12-20

AI Summary

This paper addresses continual test-time adaptation (CTTA) for text understanding under sequential exposure to unseen test domains. Existing CTTA methods suffer from an imbalance between noise filtering and historical knowledge accumulation: aggressive filtering discards useful information, while naive accumulation hinders adaptive fusion. To resolve this, we propose a novel framework that jointly optimizes error suppression and cross-domain generalization. Specifically, we introduce a domain-aware teacher model with strong generalization capability; design a “refine-filter” mechanism based on Dropout-induced consistency calibration to suppress error propagation; and develop an incremental PCA-driven dynamic cross-domain semantic accumulation strategy to enhance generalization to unknown domains. Evaluated across multiple text understanding tasks, our method significantly outperforms state-of-the-art CTTA and test-time adaptation (TTA) baselines, achieving superior robustness and generalization performance.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Computer Vision: Adversarial Attacks & Robustness

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
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Text understanding often suffers from domain shifts. To handle testing domains, domain adaptation (DA) is trained to adapt to a fixed and observed testing domain; a more challenging paradigm, test-time adaptation (TTA), cannot access the testing domain during training and online adapts to the testing samples during testing, where the samples are from a fixed domain. We aim to explore a more practical and underexplored scenario, continual test-time adaptation (CTTA) for text understanding, which involves a sequence of testing (unobserved) domains in testing. Current CTTA methods struggle in reducing error accumulation over domains and enhancing generalization to handle unobserved domains: 1) Noise-filtering reduces accumulated errors but discards useful information, and 2) accumulating historical domains enhances generalization, but it is hard to achieve adaptive accumulation. In this paper, we propose a CTTA-T (continual test-time adaptation for text understanding) framework adaptable to evolving target domains: it adopts a teacher-student framework, where the teacher is domain-aware and generalized for evolving domains. To improve teacher predictions, we propose a refine-then-filter based on dropout-driven consistency, which calibrates predictions and removes unreliable guidance. For the adaptation-generalization trade-off, we construct a domain-aware teacher by dynamically accumulating cross-domain semantics via incremental PCA, which continuously tracks domain shifts. Experiments show CTTA-T excels baselines.

Problem

• Adapts to evolving text domains sequentially
• Reduces error accumulation while preserving useful information
• Enhances generalization to unobserved domains dynamically

Innovation

• Teacher-student framework with domain-aware teacher
• Refine-then-filter method using dropout-driven consistency
• Dynamic cross-domain accumulation via incremental PCA