Saved in:
Bibliographic Details
Main Authors: Junike, Gero, Oesting, Marco
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.00490
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908624459137024
author Junike, Gero
Oesting, Marco
author_facet Junike, Gero
Oesting, Marco
contents Neural networks are able to approximate any continuous function on a compact set. However, it is not obvious how to quantify the error of the neural network, i.e., the remaining bias between the function and the neural network. Here, we propose the application of extreme value theory to quantify large values of the error, which are typically relevant in applications. The distribution of the error beyond some threshold is approximately generalized Pareto distributed. We provide a new estimator of the shape parameter of the Pareto distribution suitable to describe the error of neural networks. Numerical experiments are provided.
format Preprint
id arxiv_https___arxiv_org_abs_2511_00490
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Accuracy estimation of neural networks by extreme value theory
Junike, Gero
Oesting, Marco
Machine Learning
Probability
Neural networks are able to approximate any continuous function on a compact set. However, it is not obvious how to quantify the error of the neural network, i.e., the remaining bias between the function and the neural network. Here, we propose the application of extreme value theory to quantify large values of the error, which are typically relevant in applications. The distribution of the error beyond some threshold is approximately generalized Pareto distributed. We provide a new estimator of the shape parameter of the Pareto distribution suitable to describe the error of neural networks. Numerical experiments are provided.
title Accuracy estimation of neural networks by extreme value theory
topic Machine Learning
Probability
url https://arxiv.org/abs/2511.00490