Calibration through the Lens of Indistinguishability

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gopalan, Parikshit, Hu, Lunjia
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908515496361984
author Gopalan, Parikshit
Hu, Lunjia
author_facet Gopalan, Parikshit
Hu, Lunjia
contents Calibration is a classical notion from the forecasting literature which aims to address the question: how should predicted probabilities be interpreted? In a world where we only get to observe (discrete) outcomes, how should we evaluate a predictor that hypothesizes (continuous) probabilities over possible outcomes? The study of calibration has seen a surge of recent interest, given the ubiquity of probabilistic predictions in machine learning. This survey describes recent work on the foundational questions of how to define and measure calibration error, and what these measures mean for downstream decision makers who wish to use the predictions to make decisions. A unifying viewpoint that emerges is that of calibration as a form of indistinguishability, between the world hypothesized by the predictor and the real world (governed by nature or the Bayes optimal predictor). In this view, various calibration measures quantify the extent to which the two worlds can be told apart by certain classes of distinguishers or statistical measures.
format Preprint
id arxiv_https___arxiv_org_abs_2509_02279
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Calibration through the Lens of Indistinguishability
Gopalan, Parikshit
Hu, Lunjia
Machine Learning
Computer Science and Game Theory
Calibration is a classical notion from the forecasting literature which aims to address the question: how should predicted probabilities be interpreted? In a world where we only get to observe (discrete) outcomes, how should we evaluate a predictor that hypothesizes (continuous) probabilities over possible outcomes? The study of calibration has seen a surge of recent interest, given the ubiquity of probabilistic predictions in machine learning. This survey describes recent work on the foundational questions of how to define and measure calibration error, and what these measures mean for downstream decision makers who wish to use the predictions to make decisions. A unifying viewpoint that emerges is that of calibration as a form of indistinguishability, between the world hypothesized by the predictor and the real world (governed by nature or the Bayes optimal predictor). In this view, various calibration measures quantify the extent to which the two worlds can be told apart by certain classes of distinguishers or statistical measures.
title Calibration through the Lens of Indistinguishability
topic Machine Learning
Computer Science and Game Theory
url https://arxiv.org/abs/2509.02279