Jingsen Kong | Searcharxiv

arXiv2026

Differentially Private Conformal Prediction

Conformal prediction (CP) has attracted broad attention as a simple and flexible framework for uncertainty quantification through prediction sets. In this work, we study how to deploy CP under differential privacy (DP) in a statistically efficient manner. We first introduce differential CP, a non-splitting conformal procedure that avoids the efficiency loss caused by data splitting and serves as a bridge between oracle CP and private conformal inference. By exploiting the stability properties of DP mechanisms, differential CP establishes a direct connection to oracle CP and inherits corresponding validity behavior. Building on this idea, we develop Differentially Private Conformal Prediction (DPCP), a fully private procedure that combines DP model training with a private quantile mechanism for calibration. We establish the end-to-end privacy guarantee of DPCP and investigate its coverage properties under additional regularity conditions. We further study the efficiency of both differential CP and DPCP under empirical risk minimization and general regression models, showing that DPCP can produce tighter prediction sets than existing private split conformal approaches under the same privacy budget. Numerical experiments on synthetic and real datasets demonstrate the practical effectiveness of the proposed methods.

stat.ML↗

arXiv2025

Fair Conformal Prediction for Incomplete Covariate Data

Conformal prediction provides a distribution-free framework for uncertainty quantification. This study explores the application of conformal prediction in scenarios where covariates are missing, which introduces significant challenges for uncertainty quantification. We establish that marginal validity holds for imputed datasets across various mechanisms of missing data and most imputation methods. Building on the framework of nonexchangeable conformal prediction, we demonstrate that coverage guarantees depend on the mask. To address this, we propose a nonexchangeable conformal prediction method for missing covariates that satisfies both marginal and mask-conditional validity. However, as this method does not ensure asymptotic conditional validity, we further introduce a localized conformal prediction approach that employs a novel score function based on kernel smoothing. This method achieves marginal, mask-conditional, and asymptotic conditional validity under certain assumptions. Extensive simulation studies and real-data analysis demonstrate the advantages of these proposed methods.

stat.ME↗