Abstract:Data-driven surrogate models have shown immense potential as quick, inexpensive approximations to complex numerical and experimental modelling tasks. However, most surrogate models of physical systems do not quantify their uncertainty, rendering their predictions unreliable, requiring further validation. Though Bayesian approximations offer some solace in estimating the error associated with these models, they cannot provide guarantees, and the quality of their inferences depends on the availability of prior information and good approximations to posteriors for complex problems. This is particularly pertinent to multi-variable or spatio-temporal problems. Our work constructs and formalises a conformal prediction framework that satisfies marginal coverage for spatio-temporal predictions in a model-agnostic manner, requiring near-zero computational costs. We provide an extensive empirical study of the application of the framework to ascertain valid error bars that provide guaranteed coverage across the surrogate model's domain of operation. The application scope of our work extends across a large range of spatio-temporal models, from solving partial differential equations to weather forecasting. Through the applications, the paper looks at providing statistically valid error bars for deterministic models, as well as crafting guarantees to the error bars of probabilistic models. Our conformal prediction formalisation provides guaranteed coverage of the surrogate model, regardless of model architecture, and its training regime and is unbothered by the curse of dimensionality.

Conformal Uncertainty Sets for Robust Optimization

New Robust Optimization Approach Induced by Flexible Uncertainty Set: Optimization under Continuous Uncertainty

Bayesian Optimization with Conformal Prediction Sets

End-to-End Conformal Calibration for Optimization Under Uncertainty

Conformal Contextual Robust Optimization

Multistage Robust Mixed-Integer Optimization under Endogenous Uncertainty

Learning Decision-Focused Uncertainty Sets in Robust Optimization

Optimization under Decision-Dependent Uncertainty

A survey of decision making and optimization under uncertainty

Conformal Inverse Optimization

Outlier Detection Using Robust Optimization with Uncertainty Sets Constructed from Risk Measures.

Optimization under Connected Uncertainty

Decision-Focused Uncertainty Quantification

Robust optimal control with adjustable uncertainty sets

First order dependence on uncertainty sets in robust optimization

Online Calibrated and Conformal Prediction Improves Bayesian Optimization

Projection-based Robust Optimization with Symbolic Computation

Conformal Prediction Sets Improve Human Decision Making

Robust Optimization in High-Dimensional Data Space with Support Vector Clustering

Uncertainty Quantification of Surrogate Models using Conformal Prediction

Distributionally robust optimization with decision dependent ambiguity sets