Research map / Probabilistic and scientific modeling / Molecules and proteins
Derandomized novelty detection with FDR control via conformal e-values
Meshi Bashari, Amir Epstein, Yaniv Romano, Matteo Sesia
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 |
|---|---|---|---|
| derandomize AdaDetect with conformal e-values (K=10) | AdaDetect · Synthetic 100-dim Gaussian outliers | variability of findings | Improved |
| derandomize AdaDetect with conformal e-values (K=10) at strong signal | AdaDetect · Synthetic 100-dim Gaussian outliers | power | Improved |
| derandomize AdaDetect with conformal e-values (K=10) at weak signal | AdaDetect · Synthetic 100-dim Gaussian outliers | power | Hurt |
| increase number of analyses K from 1 to 30 (ablation) | E-AdaDetect · Synthetic 100-dim Gaussian outliers, amplitude 3.4 | variability | Improved |
| increase number of analyses K from 1 to 30 (ablation) | E-AdaDetect · Synthetic 100-dim Gaussian outliers, amplitude 3.4 | power | Improved |
| increase K at low signal (amplitude 2.8) (ablation) | E-AdaDetect · Synthetic 100-dim Gaussian outliers | power | Hurt |
| set K=1 (eBH filter only) (ablation) | E-AdaDetect · Synthetic 100-dim Gaussian outliers | power vs AdaDetect | Hurt |
| t-test data-driven model weights instead of uniform weights (ablation) | E-AdaDetect · Synthetic 100-dim Gaussian outliers sparse logistic regression | power | Improved |
| t-test data-driven weights instead of uniform weights (ablation) | E-AdaDetect · musk, shuttle, KDDCup99, credit card | number of discoveries | Improved |
| data-driven weights instead of uniform weights (ablation) | E-AdaDetect · musk, shuttle, KDDCup99, credit card | algorithmic variance | Improved |
| trimmed-average weights instead of t-test weights (ablation) | E-AdaDetect · real data sets | power | Hurt |
| derandomize one-class conformal (E-OC-Conformal) | OC-Conformal · Synthetic 100-dim Gaussian outliers | variance | Improved |
Closest papers by meaning
- Conditional Testing based on Localized Conformal $p$-values (ICLR 2025 · similarity 0.86)
- Robust Conformal Outlier Detection under Contaminated Reference Data (ICML 2025 · similarity 0.86)
- Test-time Augmentation Improves Efficiency in Conformal Prediction (CVPR 2025 · similarity 0.85)
- Conformalized Multiple Testing after Data-dependent Selection (NeurIPS 2024 · similarity 0.84)
- Robust Conformal Prediction Using Privileged Information (NeurIPS 2024 · similarity 0.84)
- Synthetic-powered predictive inference (NeurIPS 2025 · similarity 0.84)
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.
