Enregistré dans:
Détails bibliographiques
Auteurs principaux: Møllersen, Kajsa, Holsbø, Einar
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:https://arxiv.org/abs/2303.07272
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916843263885312
author Møllersen, Kajsa
Holsbø, Einar
author_facet Møllersen, Kajsa
Holsbø, Einar
contents State-of-the-art (SOTA) performance refers to the highest performance achieved by some model on a test sample, preferably under controlled conditions such as public data (reproducibility) or public challenges (independent sample). Thousands of classifiers are applied, and the highest performance becomes the new reference point for a particular problem. In effect, this set-up is an estimate of the expected best performance among all classifiers applied to a random sample; a sample maximum estimate. In this paper, we argue that SOTA should instead be estimated by the expected performance of the best classifier, which can be done without knowing which classifier it is. Our contribution is the formal distinction between the two, and an investigation into the practical consequences of using the former to estimate the latter. This is done by presenting sample maximum estimator distributions for non-identical and dependent classifiers. We illustrate the impact on real world examples from public challenges.
format Preprint
id arxiv_https___arxiv_org_abs_2303_07272
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Accounting for multiplicity in machine learning benchmark performance
Møllersen, Kajsa
Holsbø, Einar
Methodology
Machine Learning
State-of-the-art (SOTA) performance refers to the highest performance achieved by some model on a test sample, preferably under controlled conditions such as public data (reproducibility) or public challenges (independent sample). Thousands of classifiers are applied, and the highest performance becomes the new reference point for a particular problem. In effect, this set-up is an estimate of the expected best performance among all classifiers applied to a random sample; a sample maximum estimate. In this paper, we argue that SOTA should instead be estimated by the expected performance of the best classifier, which can be done without knowing which classifier it is. Our contribution is the formal distinction between the two, and an investigation into the practical consequences of using the former to estimate the latter. This is done by presenting sample maximum estimator distributions for non-identical and dependent classifiers. We illustrate the impact on real world examples from public challenges.
title Accounting for multiplicity in machine learning benchmark performance
topic Methodology
Machine Learning
url https://arxiv.org/abs/2303.07272