ResearcharXivNEW

Optimal Deterministic Multicalibration and Omniprediction

Noarov 2026-06-18
Georgy NoarovAaron Roth

A model is multicalibrated on a collection of group weights $G$ if it is calibrated -- i.e. unbiased even conditional on its prediction -- not just overall, but also after reweighting contexts by each $g \in G$. It is a useful property for many downstream applications and is a basic desideratum of trustworthy machine learning. Before this work, all predictors known to attain the minimax-optimal $\

Read on arXiv
Data aggregated and editorially reviewed by TrendMing.

Key Contributions

  • A model is multicalibrated on a collection of group weights $G$ if it is calibrated -- i.e.
  • unbiased even conditional on its prediction -- not just overall, but also after reweighting contexts by each $g \in G$.
  • It is a useful property for many downstream applications and is a basic desideratum of trustworthy machine learning.

Research Themes

AIResearch