Research map / Label-efficient and robust learning / Domain adaptation and generalization
Conformal Calibration Transfer
Achref Doula
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 predictive KL alignment term to transport learning (ablation) | TCC-KS · CIFAR-100-C, epoch 150 | delta+: 0.0521 -> 0.0152 | Improved |
| add predictive KL alignment term to transport learning (ablation) | TCC-KS · Tiny-ImageNet-C, epoch 150 | delta+: 0.1006 -> 0.0148 | Improved |
| add predictive KL alignment term to transport learning (ablation) | weighted-TCC · Tiny-ImageNet-C, epoch 150 | ESS%: 48.4 -> 91.8 | Improved |
| add predictive KL alignment term to transport learning (ablation) | Transported CP · CIFAR-100-C, epoch 150 | set size: 44.03 -> 37.18 | Improved |
| train transport longer (epoch 5 to 150) without KL (ablation) | TCC-KS · Tiny-ImageNet-C | delta+: 0.3422 -> 0.1006 | Improved |
| add KS-based inflation (TCC-KS) to Transported CP | Transported CP · CIFAR-100-C, alpha=0.1 | set size: 37.18 -> 44.35 | Hurt |
| add KS-based inflation (TCC-KS) to Transported CP | Transported CP · CIFAR-100-C stress test, alpha=0.1 | coverage: 0.88 -> 0.91 | Improved |
| use transport plus weighted-TCC instead of WCP without transport | WCP · CIFAR-100-C, alpha=0.1 | set size: 43.90 -> 37.15 | Improved |
| use transport plus weighted-TCC instead of WCP without transport | WCP · Tiny-ImageNet-C stress test, alpha=0.1 | set size: 120.26 -> 76.84 | Improved |
| TCC-KS on SAR->RGB cross-modal shift vs Transported CP | Transported CP · SAR->RGB, alpha=0.2 | coverage: 0.85 -> 0.99 | Improved |
| use TCC-KS on under-coverage cases of transport-only | Transported CP · 328 configurations | coverage: 87.9% -> 97.8% | Improved |
| add +KL on SAR->RGB at epoch 150 (ablation) | Transported CP · SAR->RGB, alpha=0.1 | coverage: 1.000 -> 0.890 | Hurt |
Closest papers by meaning
- Self-Calibrating Conformal Prediction (NeurIPS 2024 · similarity 0.86)
- Beyond One-Hot Labels: Semantic Mixing for Model Calibration (ICML 2025 · similarity 0.86)
- Non-exchangeable Conformal Prediction with Optimal Transport: Tackling Distribution Shift with Unlabeled Data (NeurIPS 2025 · similarity 0.86)
- Generalized Venn and Venn-Abers Calibration with Applications in Conformal Prediction (ICML 2025 · similarity 0.86)
- Pseudo-Calibration: Improving Predictive Uncertainty Estimation in Unsupervised Domain Adaptation (ICML 2024 · similarity 0.85)
- Robust Conditional Conformal Prediction via Branched Normalizing Flow (NeurIPS 2026 · 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.
