A probabilistic framework for online test-time adaptation

Daniel Corrales; David Ríos Insua

stat.MLJun 24, 2026

A probabilistic framework for online test-time adaptation

Authors: Daniel Corrales, David Ríos Insua

Organizations: Escuela de Doctorado, Universidad Aut´onoma de Madrid, 28049 Madrid, Spain · Institute of Mathematical Sciences, ICMAT-CSIC, 28049 Madrid, Spain

Abstract

This paper presents a probabilistic framework for online test-time adaptation problems. In them, a model is trained on labeled data but must adapt to unlabeled data at test time under the assumption that training and test distributions potentially differ, that is, there might have been a distributional shift. The framework is based on a state-space modelling architecture from which parameter learning, parameter time evolution, prior tuning, and prediction can be characterized.

Explore similar work

Dual Strategies for Test-Time Adaptation
Apr 19, 2026Nam Nguyen Phuong, Duc Nguyen The Minh, Phi Le Nguyen +2Distribution Shifts Duality
A Decision-Theoretic View of Test-Time Training: When, How Far, and Which Directions to Adapt
Jun 14, 2026Tomoya WakayamaDistribution Shifts Bayesian
When is Test-Time Adaptation Identifiable From Unlabeled Evidence?
Sep 11, 2026Kartik Jhawar, Lipo WangIncomplete Evidence

Apr 19, 2026 · cs.CVJ/K move · Enter open · S save

Dual Strategies for Test-Time Adaptation

Nam Nguyen Phuong, Duc Nguyen The Minh, Phi Le Nguyen, Ehsan Abbasnejad+1

Institute for AI Innovation and Societal Impact, Hanoi University of Science and Technology · Monash University · Adelaide University

Conventional test-time adaptation (TTA) approaches typically adapt the model using only a small fraction of test samples, often those with low-entropy predictions, thereby failing to fully leverage the available information in the test distribution. This paper introduces DualTTA, a novel framework that improves performance under distribution shifts by utilizing a larger and more diverse set of test samples. DualTTA identifies two distinct groups: one where the model's predictions are likely consistent with the underlying semantics, and another where predictions are likely incorrect. For the first group, it minimizes prediction entropy to reinforce reliable decisions; for the second, it maximizes entropy to suppress overconfident errors and unlearn spurious behavior. These groups are adaptively selected using a new reliability criterion that measures prediction stability under both semantic-preserving and semantic-altering transformations, addressing the limitations of purely entropy-based selection. We further provide theoretical analysis and empirical justification showing that our approach enables a tighter separation between reliable and unreliable samples, in the context of their suitability for adaptation, leading to provably more effective model updates.

Distribution Shifts Duality Unreliable Reasoning

Open paper Ask about it PDF

Abstract

Explore similar work

Dual Strategies for Test-Time Adaptation

A Decision-Theoretic View of Test-Time Training: When, How Far, and Which Directions to Adapt

When is Test-Time Adaptation Identifiable From Unlabeled Evidence?