A learning problem whose consistency is equivalent to the non-existence of real-valued measurable cardinals

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Pestov, Vladimir G.
Format: Preprint
Published: 2020
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911861553758208
author Pestov, Vladimir G.
author_facet Pestov, Vladimir G.
contents We show that the $k$-nearest neighbour learning rule is universally consistent in a metric space $X$ if and only if it is universally consistent in every separable subspace of $X$ and the density of $X$ is less than every real-measurable cardinal. In particular, the $k$-NN classifier is universally consistent in every metric space whose separable subspaces are sigma-finite dimensional in the sense of Nagata and Preiss if and only if there are no real-valued measurable cardinals. The latter assumption is relatively consistent with ZFC, however the consistency of the existence of such cardinals cannot be proved within ZFC. Our results were inspired by an example sketched by Cérou and Guyader in 2006 at an intuitive level of rigour.
format Preprint
id arxiv_https___arxiv_org_abs_2005_01886
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle A learning problem whose consistency is equivalent to the non-existence of real-valued measurable cardinals
Pestov, Vladimir G.
Machine Learning
Logic
62H30, 54F45, 03E55
I.2.6
We show that the $k$-nearest neighbour learning rule is universally consistent in a metric space $X$ if and only if it is universally consistent in every separable subspace of $X$ and the density of $X$ is less than every real-measurable cardinal. In particular, the $k$-NN classifier is universally consistent in every metric space whose separable subspaces are sigma-finite dimensional in the sense of Nagata and Preiss if and only if there are no real-valued measurable cardinals. The latter assumption is relatively consistent with ZFC, however the consistency of the existence of such cardinals cannot be proved within ZFC. Our results were inspired by an example sketched by Cérou and Guyader in 2006 at an intuitive level of rigour.
title A learning problem whose consistency is equivalent to the non-existence of real-valued measurable cardinals
topic Machine Learning
Logic
62H30, 54F45, 03E55
I.2.6
url https://arxiv.org/abs/2005.01886