Research map / Learning theory and algorithms / Prediction and decision losses
Practical Adversarial Multivalid Conformal Prediction
Osbert Bastani, Varun Gupta, Christopher Jung, Georgy Noarov, Ramya Ramalingam, Aaron Roth
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 |
|---|---|---|---|
| Use MVP rather than RAPS | ImageNet classification with ResNet-152 · ImageNet validation | marginal coverage: 0.90523 -> 0.902133 | No clear effect |
| Use MVP rather than RAPS | ImageNet classification with ResNet-152 · ImageNet validation | average prediction-set size: 2.0506 -> 2.13986 | Hurt |
| Use MVP instead of split conformal while training the regression model on all past observations | Synthetic linear regression · Synthetic linear regression, marginal coverage | empirical coverage | Improved |
| Use MVP instead of split conformal while training the regression model on all past observations | Synthetic linear regression · Synthetic linear regression, marginal coverage | average interval width | Improved |
| Use MVP rather than naive split conformal for intersecting-group calibration | Synthetic regression with 20 intersecting groups · Synthetic regression, group-wise coverage | group-wise coverage | Improved |
| Use MVP rather than conservative groupwise conformal | Synthetic regression with 20 intersecting groups · Synthetic regression, group-wise coverage | average interval width | Improved |
| Use MVP rather than split conformal under unknown state-to-state shift | Folktables income prediction · California to Pennsylvania covariate shift | empirical coverage | Improved |
| Use MVP rather than ACI for intersecting subgroups | GARCH stock-volatility prediction · Noisy stock-return groups | group-wise coverage | Improved |
| Use threshold-calibrated MVP rather than marginal-only ACI on monotonically increasing scores | Stock-return adversarial ordering · Scores increasing from 0.0 to 0.5 | average interval width: 1.839 -> 0.526 | Improved |
| Use normalized rather than un-normalized surrogate score function (ablation) | MVP · Experiments across reported settings | performance | No clear effect |
Closest papers by meaning
- Verifiably Robust Conformal Prediction (NeurIPS 2024 · similarity 0.89)
- Robust Yet Efficient Conformal Prediction Sets (ICML 2024 · similarity 0.88)
- Provably Robust Conformal Prediction with Improved Efficiency (ICLR 2024 · similarity 0.88)
- Robust Conformal Prediction with a Single Binary Certificate (ICLR 2025 · similarity 0.87)
- Efficient Robust Conformal Prediction via Lipschitz-Bounded Networks (ICML 2025 · similarity 0.87)
- The Pitfalls and Promise of Conformal Inference Under Adversarial Attacks (ICML 2024 · similarity 0.87)
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.
