Research map / Label-efficient and robust learning / Test-time adaptation
Tent: Fully Test-Time Adaptation by Entropy Minimization
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 |
|---|---|---|---|
| minimize prediction entropy while adapting normalization statistics and channel-wise affine parameters | Tent · CIFAR-10-C, severity 5 | error (%): 15.7 -> 14.3 | Helped |
| minimize prediction entropy while adapting normalization statistics and channel-wise affine parameters | Tent · CIFAR-100-C, severity 5 | error (%): 41.2 -> 37.3 | Helped |
| use online test-time adaptation rather than ANT training | ImageNet-C · ImageNet-C | error (%): 50.2% -> 44.0% | Helped |
| adapt offline rather than online | Tent on ImageNet-C · ImageNet-C | error (%): 44.0% -> 42.3% | Helped |
| adapt on SVHN-to-MNIST | Tent · SVHN to MNIST | error (%): 11.1 -> 8.2 | Helped |
| adapt on SVHN-to-MNIST-M | Tent · SVHN to MNIST-M | error (%): 22.2 -> 36.8 | Hurt |
| adapt on SVHN-to-USPS | Tent · SVHN to USPS | error (%): 18.4 -> 14.4 | Helped |
| adapt to GTA-to-Cityscapes target data | Tent · GTA to Cityscapes | intersection-over-union (%): 31.4% -> 35.8% | Helped |
| adapt all layers except the final classifier instead of the default modulation parameters (ablation) | Tent on VisDA-C · VisDA-C | error (%): 45.6% -> 39.6% | Helped |
| remove updating normalization statistics (ablation) | Tent feature modulation · corruption robustness | error | Hurt |
| remove updates to affine transformation parameters (ablation) | Tent feature modulation · corruption robustness | error | Hurt |
| update only the last layer (ablation) | Tent · corruption robustness | error | Mixed |
Closest papers by meaning
- Ranked Entropy Minimization for Continual Test-Time Adaptation (ICML 2025 · similarity 0.88)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (NeurIPS 2026 · similarity 0.87)
- Unified Entropy Optimization for Open-Set Test-Time Adaptation (CVPR 2024 · similarity 0.86)
- ReservoirTTA: Prolonged Test-time Adaptation for Evolving and Recurring Domains (NeurIPS 2025 · similarity 0.86)
- Frustratingly Easy Test-Time Adaptation of Vision-Language Models (NeurIPS 2024 · similarity 0.85)
- Multi-Label Test-Time Adaptation with Bound Entropy Minimization (ICLR 2025 · similarity 0.85)
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.