Multivariate Conformal Prediction via Conformalized Gaussian Scoring

Sacha Braun, Eugène Berta, Michael I. Jordan, Francis Bach · 2025

Read Multivariate Conformal Prediction via Conformalized Gaussian Scoring by Sacha Braun, Eugène Berta, Michael I. Jordan, Francis Bach (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-20941.txt.

Description

Sacha Braun and co-authors carry a variance-normalizing trick for prediction bands into many outputs at once, in this 2025 paper. Conformal prediction wraps a model's guesses in sets with guaranteed coverage, but plain scores under-cover where noise swells. The fix whitens residuals with a learned local covariance, the many-output echo of dividing by local spread.

The authors prove when this yields coverage that adapts to each case, not just on average. The scoring rule is a closed-form distance, cheap to compute, and it handles missing outputs and partial reveals. Tests on synthetic and real data beat earlier methods.

Catalog record

CREATORSacha Braun, Eugène Berta, Michael I. Jordan, Francis Bach
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-20941

Browse the full library · About the Tower of Babel