AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

CalPro: Prior-Aware Evidential Conformal Prediction with Structure-Aware Sensitivity Bounds for Protein Structures

Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma

ICML 2026 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Conformal prediction methods

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 changedWhereMeasuredEffect
add split-conformal calibration (ablation)CalPro · Protein benchmarkscoverage (%): 78.3 -> 90.1Improved
add NIG evidential head (ablation)CalPro · Protein benchmarkssharpness (Angstrom): 1.98 -> 1.42Improved
add prior regularizer L_prior (vs priors as features only) (ablation)CalPro · Protein benchmarksSpearman correlation with error: 0.76 -> 0.83Improved
add bio-prior inputs and regularizer (ablation)CalPro · Protein benchmarksSpearman correlation with error: 0.74 -> 0.83Improved
add priors in modality shift (ablation)CalPro · X-ray to cryo-EMcoverage at 90% nominal (%): 84.7 -> 86.2Improved
add differentiable surrogate loss L_soft-conf (ablation)CalPro · Protein benchmarksSpearman correlation with error: 0.81 -> 0.83Improved
use GNN head instead of MLP head (ablation)CalPro · Protein benchmarksSpearman correlation with error: 0.78 -> 0.83Improved
add prior regularizer: constant reduction of Ls and Ms (ablation)CalPro · Shift-sensitivity boundM_s L_s epsilon (eps=0.1): 1.33 -> 0.68 -> 0.41Improved
invert the priors (adversarial) (ablation)CalPro · X-ray to cryo-EMcoverage at 90% nominal (%): 86.2 -> 82.8Hurt
shuffle the priors (ablation)CalPro · X-ray to cryo-EMcoverage at 90% nominal (%): 86.2 -> 84.1Hurt
train CQR on CalPro backbone with adapted prior loss (ablation)CQR · X-ray to cryo-EMcoverage at 90% nominal (%): 83.7 -> 84.3Mixed
use CalPro instead of CQR for filtering docking posesdocking filter · PDBbinddocking success (%): 70.1 -> 74.8Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy