When Does Trimming Help Conformal Prediction? A Retained-Law Diagnostic under Calibration Contamination
Abstract
Trimming suspicious calibration points is a common response to contamination in conformal prediction. Its effect on clean-target coverage, however, is governed by the retained law induced by trimming, not by the contamination level alone. We analyse fixed-threshold trimming as conditioning rather than purification. It replaces the contaminated calibration law with a retained law, reducing clean-target coverage to a one-dimensional score-CDF transfer problem with an exact finite-sample identity. A componentwise bound on the transfer gap gives a population-level diagnostic. This separates a clean-side covariance cost from a retained-contamination cost, governed by the dirty-to-clean retention ratio. Trimming helps when the anomaly score separates retention probabilities while remaining score-neutral on the clean population. Otherwise, it cannot substantially reduce contamination through the retained mixture coefficient. We also give finite-sample certificate templates that provide numerical guarantees under independent audit.
Related
- Multi-Variable Conformal Prediction: Optimizing Prediction Sets using Multiple Calibration Variables
- When does conformal calibration need censoring weights? Cause-of-failure prediction sets under competing risks
- Sketched Calibration for Conformal Prediction under Covariate Shift
- Particle-Based Conformal Prediction for Contact-Aware Uncertainty Calibration in Stratified Configuration Spaces
- Abstention Spends Its Budget on Local Label Variance More Than on Novelty: Auditing Conformal Calibration and Selective Prediction under Molecular Scaffold Shift
BibTeX
@misc{wang2026when,
title = {{When Does Trimming Help Conformal Prediction? A Retained-Law Diagnostic under Calibration Contamination}},
author = {Congye Wang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2605.06204},
eprint = {2605.06204},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.06204},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - When Does Trimming Help Conformal Prediction? A Retained-Law Diagnostic under Calibration Contamination
AU - Congye Wang
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2605.06204
UR - https://arxiv.org/abs/2605.06204
AB - Trimming suspicious calibration points is a common response to contamination in conformal prediction. Its effect on clean-target coverage, however, is governed by the retained law induced by trimming, not by the contamination level alone. We analyse fixed-threshold trimming as conditioning rather than purification. It replaces the contaminated calibration law with a retained law, reducing clean-target coverage to a one-dimensional score-CDF transfer problem with an exact finite-sample identity. A componentwise bound on the transfer gap gives a population-level diagnostic. This separates a clean-side covariance cost from a retained-contamination cost, governed by the dirty-to-clean retention ratio. Trimming helps when the anomaly score separates retention probabilities while remaining score-neutral on the clean population. Otherwise, it cannot substantially reduce contamination through the retained mixture coefficient. We also give finite-sample certificate templates that provide numerical guarantees under independent audit.
ER -