Strategic Conformal Prediction

TL;DR

This paper introduces Strategic Conformal Prediction, a novel framework for robust uncertainty quantification in machine learning models that face strategic manipulations, with strong theoretical guarantees and validated experimental performance.

Contribution

It presents a new approach to conformal prediction that remains reliable under strategic environment alterations, addressing a key gap in existing uncertainty quantification methods.

Findings

Achieves marginal and training-conditional coverage guarantees

Demonstrates robustness against arbitrary strategic manipulations

Outperforms existing methods in experimental evaluations

Abstract

When a machine learning model is deployed, its predictions can alter its environment, as better informed agents strategize to suit their own interests. With such alterations in mind, existing approaches to uncertainty quantification break. In this work we propose a new framework, Strategic Conformal Prediction, which is capable of robust uncertainty quantification in such a setting. Strategic Conformal Prediction is backed by a series of theoretical guarantees spanning marginal coverage, training-conditional coverage, tightness and robustness to misspecification that hold in a distribution-free manner. Experimental analysis further validates our method, showing its remarkable effectiveness in face of arbitrary strategic alterations, whereas other methods break.