Research map / Learning theory and algorithms / Prediction and decision losses
Few-Shot Calibration of Set Predictors via Meta-Learned Cross-Validation-Based Conformal Prediction
Sangwoo Park, Kfir M. Cohen, Osvaldo Simeone
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 meta-learning across tasks to cross-validation CP (meta-XB vs XB-CP) | XB-CP · Synthetic multinomial (T=100 tasks) | average prediction set size: larger than 4 -> 3 | Improved |
| add adaptive nonconformity scores (ablation) | meta-XB · Synthetic multinomial | conditional validity vs inefficiency | Mixed |
| use cross-validation-based CP instead of validation-based CP | set predictors · Synthetic multinomial | conditional coverage | Improved |
| increase per-task examples N to 80 or more (ablation) | meta-learned set predictors · Synthetic multinomial | inefficiency | Hurt |
| add meta-learning in few-example regime (N up to 40) (ablation) | VB-CP and XB-CP · Synthetic multinomial | inefficiency | Improved |
| use meta-VB with very many meta-training tasks (T above about 10^4) (ablation) | meta-VB · Synthetic multinomial | per-task validity | Hurt |
| add meta-learning to conformal predictors | VB-CP and XB-CP · RadioML 2018.01A modulation classification | prediction set size | Improved |
| use meta-XB instead of meta-VB | meta-learned CP · RadioML 2018.01A modulation classification | per-task coverage validity | Improved |
| use meta-XB instead of meta-VB | meta-learned CP · miniImageNet binary, N=4, alpha=0.2 | per-task coverage validity | Improved |
Closest papers by meaning
- Split conformal classification with unsupervised calibration (NeurIPS 2025 · similarity 0.86)
- Valid Selection among Conformal Sets (NeurIPS 2025 · similarity 0.86)
- Self-Calibrating Conformal Prediction (NeurIPS 2024 · similarity 0.86)
- Direct Prediction Set Minimization via Bilevel Conformal Classifier Training (ICML 2025 · similarity 0.86)
- Provably Reliable Conformal Prediction Sets in the Presence of Data Poisoning (ICLR 2025 · similarity 0.85)
- Backward Conformal Prediction (NeurIPS 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.
