Research map / Label-efficient and robust learning / Test-time adaptation
MEMO: Test Time Robustness via Adaptation and Augmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 MEMO test-time adaptation | ResNet-50 · ImageNet-C | mCE: 76.7 -> 69.9 | Helped |
| Add MEMO on top of DeepAugment+AugMix model | ResNet-50 · ImageNet-C | mCE: 53.6 -> 49.8 | Helped |
| Add MEMO | RVT*-small · ImageNet-R | error (%): 52.3 -> 43.8 | Helped |
| Add MEMO | ResNet-26 · CIFAR-10-C | average error (%): 22.5 -> 19.6 | Helped |
| Add MEMO to ResNet-50 | ResNet-50 · ImageNet-A | error (%): 100.0 -> 99.1 | Helped |
| Use TTA only (no adaptation) on ImageNet-C | ResNet-50 · ImageNet-C | mCE: 76.7 -> 77.9 | Hurt |
| Use TTA instead of MEMO on ImageNet-A | RVT*-small · ImageNet-A | error (%): 73.9 -> 68.9 | Helped |
| Add MEMO to WSL ResNext-101 | ResNext-101 WSL · ImageNet-A | error (%): 54.9 -> 43.2 | Helped |
| Replace marginal entropy with pairwise cross entropy (ablation) | MEMO · CIFAR-10-C | average error (%): 19.6 -> 20.0 | Hurt |
| Replace marginal entropy with conditional entropy (ablation) | MEMO · ImageNet-C | mCE: 40.6 -> 41.2 | Hurt |
| Replace AugMix with standard augmentations (ablation) | MEMO · CIFAR-10-C | average error (%): 19.6 -> 20.2 | Hurt |
| Replace AugMix with standard augmentations (ablation) | MEMO · CIFAR-10 | error (%): 7.3 -> 7.2 | No clear effect |
Closest papers by meaning
- NC-TTT: A Noise Constrastive Approach for Test-Time Training (CVPR 2024 · similarity 0.87)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (NeurIPS 2026 · similarity 0.87)
- Entropy is not Enough for Test-Time Adaptation: From the Perspective of Disentangled Factors (ICLR 2024 · similarity 0.86)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · similarity 0.86)
- Robust Test-Time Adaptation for Zero-Shot Prompt Tuning (AAAI 2024 · similarity 0.86)
- Test-Time Model Adaptation with Only Forward Passes (ICML 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.