Score: 1

Sample-Efficient Omniprediction for Proper Losses

Published: October 14, 2025 | arXiv ID: 2510.12769v1

By: Isaac Gibbs, Ryan J. Tibshirani

BigTech Affiliations: University of California, Berkeley

Potential Business Impact:

Helps computers make better choices for everyone.

Business Areas:
Predictive Analytics Artificial Intelligence, Data and Analytics, Software

We consider the problem of constructing probabilistic predictions that lead to accurate decisions when employed by downstream users to inform actions. For a single decision maker, designing an optimal predictor is equivalent to minimizing a proper loss function corresponding to the negative utility of that individual. For multiple decision makers, our problem can be viewed as a variant of omniprediction in which the goal is to design a single predictor that simultaneously minimizes multiple losses. Existing algorithms for achieving omniprediction broadly fall into two categories: 1) boosting methods that optimize other auxiliary targets such as multicalibration and obtain omniprediction as a corollary, and 2) adversarial two-player game based approaches that estimate and respond to the ``worst-case" loss in an online fashion. We give lower bounds demonstrating that multicalibration is a strictly more difficult problem than omniprediction and thus the former approach must incur suboptimal sample complexity. For the latter approach, we discuss how these ideas can be used to obtain a sample-efficient algorithm through an online-to-batch conversion. This conversion has the downside of returning a complex, randomized predictor. We improve on this method by designing a more direct, unrandomized algorithm that exploits structural elements of the set of proper losses.

Country of Origin
🇺🇸 United States

Page Count
37 pages

Category
Computer Science:
Machine Learning (CS)