Guaranteed prediction sets for functional surrogate models

TL;DR

This paper introduces a method to create statistically guaranteed prediction sets for functional surrogate models, enhancing reliability in PDE emulation by leveraging low-dimensional representations and set-propagation techniques.

Contribution

It presents a novel, model-agnostic approach for constructing guaranteed prediction sets for complex functional models using zonotopes and SVD-based error representation.

Findings

Provides conformal prediction coverage guarantees

Applicable to neural operators and other complex models

Ensures reliable uncertainty quantification in PDE emulation

Abstract

We propose a method for obtaining statistically guaranteed prediction sets for functional machine learning methods: surrogate models which map between function spaces, motivated by the need to build reliable PDE emulators. The method constructs nested prediction sets on a low-dimensional representation (an SVD) of the surrogate model's error, and then maps these sets to the prediction space using set-propagation techniques. This results in prediction sets for functional surrogate models with conformal prediction coverage guarantees. We use zonotopes as basis of the set construction, which allow an exact linear propagation and are closed under Cartesian products, making them well-suited to this high-dimensional problem. The method is model agnostic and can thus be applied to complex Sci-ML models, including Neural Operators, but also in simpler settings. We also introduce a technique to capture the truncation error of the SVD, preserving the guarantees of the method.