Research map / Label-efficient and robust learning / Test-time adaptation
STAMP: Outlier-Aware Test-Time Adaptation with Stable Memory Replay
Yu Yongcan, Lijun Sheng, Ran He, Jian Liang
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 reliable class-balanced memory (ablation) | STAMP · CIFAR100-C + SVHN-C | ACC (%): 59.3 -> 63.7 | Improved |
| add self-weighted entropy minimization (ablation) | STAMP · CIFAR100-C + SVHN-C | AUC (%): 75.1 -> 82.1 | Improved |
| add test-time step size decay alone (without SAM) (ablation) | entropy minimization · CIFAR100-C + SVHN-C | AUC (%): ACC 54.4 -> 58.9; AUC 70.4 -> 65.3 | Mixed |
| combine SAM with step size decay (ablation) | entropy minimization · CIFAR100-C + SVHN-C | H-score (%): 61.2 -> 64.4 | Improved |
| use adaptive weights with gradient backprop instead of static or EATA weights | STAMP entropy minimization · CIFAR100-C + SVHN-C | H-score | Improved |
| apply Tent entropy minimization to all samples under outliers | Tent · CIFAR10-C with Noise outliers | AUC (%): 70.4 -> 48.7 | Hurt |
| use STAMP versus best prior TTA method | CIFAR10-C outlier-aware · CIFAR10-C | AUC (%): +5.1% | Improved |
| use STAMP in closed-set setting without outliers | CIFAR10-C and CIFAR100-C · closed-set TTA | accuracy: +2.9% average over SoTTA | Improved |
| vary outlier proportion from 5% to 50% (ablation) | STAMP · CIFAR100-C | H-score (%): fluctuation within 3% | No clear effect |
| average predictions over 16 augmented views | STAMP · CIFAR10-C | outlier detection | Improved |
Later papers that compared with this method
3 later accepted papers report STAMP (base) as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Bid Farewell to Seesaw: Towards Accurate Long-Tail Session-Based Recommendation via Dual Constraints of Hybrid Intents (AAAI 2026 · better on 4 of 4 shared results)
- Tail Cues, Principal Corrections: Plug-and-Play Rectification for Open-Set Test-Time Adaptation (NeurIPS 2026 · better on 1 of 1 shared result)
- Back to Source: Open-Set Continual Test-Time Adaptation via Domain Compensation (CVPR 2026 · 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 2 of 2 on most shared results.
- SoTTA — SoTTA: Robust Test-Time Adaptation on Noisy Data Streams (NeurIPS 2023) · better on 1 of 1 shared result
- Tent — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · better on 1 of 1 shared result
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
- Unified Entropy Optimization for Open-Set Test-Time Adaptation (CVPR 2024 · similarity 0.87)
- Protected Test-Time Adaptation via Online Entropy Matching: A Betting Approach (NeurIPS 2024 · similarity 0.86)
- Active Test-Time Adaptation: Theoretical Analyses and An Algorithm (ICLR 2024 · similarity 0.86)
- AETTA: Label-Free Accuracy Estimation for Test-Time Adaptation (CVPR 2024 · similarity 0.86)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · similarity 0.86)
- Towards Real-World Test-Time Adaptation: Tri-net Self-Training with Balanced Normalization (AAAI 2024 · similarity 0.86)
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.
