# When is Test-Time Adaptation Identifiable From Unlabeled Evidence?

arXiv: https://arxiv.org/abs/2609.11235 · HTML: https://mlsanity.com/paper/2609.11235/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 31 references that are in the map.
- First posted on arXiv 10 Sep 2026.

## Builds on (most cited first, 22 of 22)

- [Benchmarking Neural Network Robustness to Common Corruptions and Perturbations](https://mlsanity.com/paper/1903.12261/) (2019)
- [Moment Matching for Multi-Source Domain Adaptation](https://mlsanity.com/paper/1812.01754/) (2018)
- [Efficient Test-Time Model Adaptation without Forgetting](https://mlsanity.com/paper/2204.02610/) (2022)
- [Towards Stable Test-Time Adaptation in Dynamic Wild World](https://mlsanity.com/paper/2302.12400/) (2023)
- [NOTE: Robust Continual Test-time Adaptation Against Temporal Correlation](https://mlsanity.com/paper/2208.05117/) (2022)
- [Leveraging Unlabeled Data to Predict Out-of-Distribution Performance](https://mlsanity.com/paper/2201.04234/) (2022)
- [Entropy is not Enough for Test-Time Adaptation: From the Perspective of Disentangled Factors](https://mlsanity.com/paper/2403.07366/) (2024)
- [TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptation](https://mlsanity.com/paper/2302.05155/) (2023)
- [Universal Test-time Adaptation through Weight Ensembling, Diversity Weighting, and Prior Correction](https://mlsanity.com/paper/2306.00650/) (2023)
- [On Pitfalls of Test-Time Adaptation](https://mlsanity.com/paper/2306.03536/) (2023)
- [Estimating Optimal Treatment Rules with an Instrumental Variable: A Partial Identification Learning Approach](https://mlsanity.com/paper/2002.02579/) (2020)
- [Partial Identifiability for Domain Adaptation](https://mlsanity.com/paper/2306.06510/) (2023)
- [Test-Time Adaptation Induces Stronger Accuracy and Agreement-on-the-Line](https://mlsanity.com/paper/2310.04941/) (2023)
- [Limits of Model Selection under Transfer Learning](https://mlsanity.com/paper/2305.00152/) (2023)
- [AETTA: Label-Free Accuracy Estimation for Test-Time Adaptation](https://mlsanity.com/paper/2404.01351/) (2024)
- [NEO: No-Optimization Test-Time Adaptation through Latent Re-Centering](https://mlsanity.com/paper/2510.05635/) (2025)
- [Optimal Decision Rules when Payoffs are Partially Identified](https://mlsanity.com/paper/2204.11748/) (2022)
- [A General Representation-Based Approach to Multi-Source Domain Adaptation](https://mlsanity.com/paper/2604.23790/) (2026)
- [On the Learnability of Test-Time Adaptation: A Recovery Complexity Perspective](https://mlsanity.com/paper/2605.28057/) (2026)
- [On the Hardness of Unsupervised Domain Adaptation: Optimal Learners and Information-Theoretic Perspective](https://mlsanity.com/paper/2507.06552/) (2025)
- [Tempora: Characterising the Time-Contingent Utility of Online Test-Time Adaptation](https://mlsanity.com/paper/2602.06136/) (2026)
- [Multi-Hypothesis Test-Time Adaptation to Mitigate Underspecification](https://mlsanity.com/paper/2607.00259/) (2026)

## Built on by (most cited first, 1 of 1)

- [When Should an In-Context Learner Expand Its Hypothesis Space?](https://mlsanity.com/paper/2610.09471/) (2026)

## Abstract (arXiv)

Test-time adaptation (TTA) offers many ways to update a deployed model without labels, but choosing the wrong update can make a strong source model worse. Recent methods therefore try to predict which adaptation will work from unlabeled test data. We ask a prior question: does the evidence given to the selector contain enough information to determine the best action at all? We show that this is not guaranteed, even with a perfect selector. If an observation channel makes two deployments look the same while their TTA rankings differ, reliable selection is impossible from that channel; richer evidence can restore the decision only when it resolves the relevant ambiguity. We make this boundary exact in a finite-batch Gaussian TTA model, where doing nothing beats mean recentering for small shifts, recentering wins beyond a unique critical shift, and the boundary shrinks as 1/√ n. Public benchmark studies on CIFAR-100-C and DomainNet-126 show the same failure mode with modern TTA methods: changing only deployment structure can reverse the oracle action while global order-blind evidence remains unchanged. The result is a practical way to separate two failure modes that are usually mixed together: a weak selector versus an information channel that cannot support the desired decision in the first place.