Minimum Volume Conformal Sets for Multivariate Regression

Conformal prediction provides a principled framework for constructing predictive sets with finite-sample validity. While much of the focus has been on univariate response variables, existing multivariate methods either impose rigid geometric assumptions or rely on flexible but computationally expensive approaches that do not explicitly optimize prediction set volume. We propose an optimization-driven framework based on a novel loss function that directly learns minimum-volume covering sets while ensuring valid coverage. This formulation naturally induces a new nonconformity score for conformal prediction, which adapts to the residual distribution and covariates. Our approach optimizes over prediction sets defined by arbitrary norm balls, including single and multi-norm formulations. Additionally, by jointly optimizing both the predictive model and predictive uncertainty, we obtain prediction sets that are tight, informative, and computationally efficient, as demonstrated in our experiments on real-world datasets.

💡 Research Summary

This paper introduces a novel optimization‑driven framework for constructing multivariate regression prediction sets that are provably valid in finite samples while directly minimizing their volume. Traditional conformal prediction methods for multivariate responses either impose rigid geometric structures (e.g., hyper‑rectangles, ellipsoids) or rely on flexible but computationally intensive density, copula, or optimal‑transport approaches that do not explicitly control set size. The authors address these shortcomings by formulating the problem as a minimum‑volume covering set (MVCS) optimization over arbitrary norm balls. A prediction set is defined as

\