Research map / Label-efficient and robust learning / Test-time adaptation
SoTTA: Robust Test-Time Adaptation on Noisy Data Streams
Taesik Gong, Yewon Kim, Taeckyung Lee, Sorn Chottananurak, Sung-Ju Lee
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add entropy-sharpness minimization (ESM) to HUS (ablation) | SoTTA · CIFAR10-C avg over scenarios | accuracy (%): 57.9 -> 81.9 | Improved |
| add HUS to ESM (ablation) | SoTTA · CIFAR10-C Noise | accuracy (%): 60.5 -> 80.0 | Improved |
| add uniform-class sampling to ESM (ablation) | SoTTA · CIFAR10-C avg over scenarios | accuracy (%): 75.4 -> 79.6 | Improved |
| add high-confidence sampling to UC + ESM (full HUS) (ablation) | SoTTA · CIFAR10-C avg over scenarios | accuracy (%): 79.6 -> 81.9 | Improved |
| high-confidence sampling only, no class balancing or ESM (ablation) | SoTTA · CIFAR10-C avg over scenarios | accuracy (%): 57.7 -> 20.0 | Hurt |
| add high-confidence sampling to uniform-class sampling without ESM (ablation) | SoTTA · CIFAR10-C Attack | accuracy (%): 70.0 -> 40.9 | Hurt |
| replace entropy minimization with ESM, no sampling (ablation) | SoTTA · CIFAR10-C Benign | accuracy (%): 69.8 -> 82.6 | Improved |
| SoTTA vs best baseline under Noise | SoTTA · CIFAR10-C Noise | accuracy (%): 73.6 -> 80.0 | Improved |
| SoTTA vs best baseline without noisy samples | SoTTA · CIFAR10-C Benign | accuracy (%): 82.4 -> 82.2 | No clear effect |
| mix noisy samples into test stream | TENT · CIFAR10-C | accuracy (%): 81.0 -> 52.1 | Hurt |
| mix noisy samples into test stream | SoTTA · CIFAR10-C | accuracy (%): 82.2 -> 80.0 | Hurt |
| increase noisy samples from 5k to 20k (ablation) | SoTTA · CIFAR10-C Noise | accuracy (%): 1.9%p degradation | No clear effect |
Later papers that compared with this method
5 later accepted papers report SoTTA as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- TiME: Test-Time Mixture-of-Experts Routing via Asymmetric CO-Optimal Transport for Continual Test-Time Adaptation (ICML 2026 · better on 2 of 2 shared results)
- Test-Time Adaptation on Noisy Data via Model-Pruning-Based Filtering and Flatness-Aware Entropy Minimization (AAAI 2025 · better on 5 of 5 shared results)
- Test-Time Adaptation with Binary Feedback (ICML 2025 · better on 1 of 1 shared result)
- Noisy Test-Time Adaptation in Vision-Language Models (ICLR 2025 · better on 1 of 1 shared result)
- STAMP: Outlier-Aware Test-Time Adaptation with Stable Memory Replay (ECCV 2024 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TENT — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AETTA: Label-Free Accuracy Estimation for Test-Time Adaptation (CVPR 2024 · similarity 0.90)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · similarity 0.89)
- Evaluation of Test-Time Adaptation Under Computational Time Constraints (ICML 2024 · similarity 0.89)
- Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment (NeurIPS 2025 · similarity 0.88)
- Persistent Test-time Adaptation in Recurring Testing Scenarios (NeurIPS 2024 · similarity 0.88)
- Bayesian Weight Enhancement with Steady-State Adaptation for Test-time Adaptation in Dynamic Environments (ICML 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
