A probabilistic framework for online test-time adaptation
Daniel Corrales, David Ríos Insua
stat.ML, cs.LG
Submitted: 2026-06-24
License: http://creativecommons.org/licenses/by-sa/4.0/
The gist: This paper presents a probabilistic framework for online test-time adaptation problems.
Terminology
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.
Sources
- A unifying framework for generalised Bayesian online learning in non-stationary environments
- LoRA-TTT: Low-Rank Test-Time Training for Vision-Language Models
- Temporal Test-Time Adaptation with State-Space Models
- Beyond Model Adaptation at Test Time: A Survey
Related papers
- Behavior of prediction performance metrics with rare events
- Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
- A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
- Scalable Posterior Uncertainty for Flexible Density-Based Clustering
- A Transport Geometry of Benign Overfitting
- Efficient quadratic entropy with distance sketches