When Does Trimming Help Conformal Prediction? A Retained-Law Diagnostic under Calibration Contamination

TL;DR

This paper analyzes how trimming suspicious calibration points affects conformal prediction coverage, providing a diagnostic tool to determine when trimming improves calibration under contamination.

Contribution

It introduces a fixed-threshold trimming analysis as conditioning, deriving a finite-sample diagnostic and bounds to assess trimming effectiveness in contaminated settings.

Findings

Trimming helps when the anomaly score separates retention probabilities without affecting clean data.

A retained law transfer problem simplifies the analysis of trimming effects.

Finite-sample certificates offer numerical guarantees under independent audits.

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.