Research map / Label-efficient and robust learning / Test-time adaptation
Test Time Adaptation via Conjugate Pseudo-labels
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 temperature scaling to softmax-entropy (conjugate PL for cross-entropy) (ablation) | ENT on cross-entropy classifier · CIFAR-100-C | error (%): 41.15 -> 36.10 | Helped |
| Add temperature scaling to softmax-entropy (ablation) | ENT on cross-entropy classifier · ImageNet-C | error (%): 45.82 -> 45.36 | Helped |
| Add temperature scaling to softmax-entropy (ablation) | ENT on cross-entropy classifier · CIFAR-10-C | error (%): 13.07 -> 12.51 | Helped |
| Replace entropy with conjugate pseudo-labels for PolyLoss classifier | PolyLoss ResNet-26 TTA · CIFAR-100-C | error (%): 38.23 -> 36.83 | Helped |
| Replace entropy with conjugate pseudo-labels for PolyLoss classifier | PolyLoss ResNet-50 TTA · ImageNet-C | error (%): 45.50 -> 44.01 | Helped |
| Replace entropy with conjugate pseudo-labels for PolyLoss classifier | PolyLoss ResNet-18 TTA · SVHN to MNIST | target error (%): 14.28 -> 10.73 | Helped |
| Replace softmax pseudo-labels with conjugate pseudo-labels for squared-loss classifier | Squared-loss ResNet-26 TTA · CIFAR-10-C | error (%): 31.81 -> 12.87 | Helped |
| Use MEMO or robust PL instead of temperature-tuned entropy minimization | Cross-entropy classifier TTA · CIFAR-100-C | error (%): 36.10 -> 37.38 | Hurt |
| Meta-learn a TTA loss instead of using TENT entropy | Cross-entropy ResNet-26 · CIFAR-10-C | error (%): 13.14 -> 12.35 | Helped |
| Use a fitted scaled softmax-entropy instead of the meta-learned loss (ablation) | Meta-TTA loss · CIFAR-10-C | error (%): 12.35 -> 12.32 | No clear effect |
| Train source with PolyLoss instead of cross-entropy, adapt with temperature-scaled entropy | ResNet-50 TTA · ImageNet-C | error (%): 45.34 -> 45.5 | Hurt |
Closest papers by meaning
- L-TTA: Lightweight Test-Time Adaptation Using a Versatile Stem Layer (NeurIPS 2024 · similarity 0.91)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · similarity 0.89)
- Unified Entropy Optimization for Open-Set Test-Time Adaptation (CVPR 2024 · similarity 0.89)
- Test-time Correlation Alignment (ICML 2025 · similarity 0.89)
- Evaluation of Test-Time Adaptation Under Computational Time Constraints (ICML 2024 · similarity 0.89)
- Persistent Test-time Adaptation in Recurring Testing Scenarios (NeurIPS 2024 · 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.