Saved in:
Bibliographic Details
Main Authors: Gillis, H. Martin, Xu, Isaac, Trappenberg, Thomas
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.08846
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908531482951680
author Gillis, H. Martin
Xu, Isaac
Trappenberg, Thomas
author_facet Gillis, H. Martin
Xu, Isaac
Trappenberg, Thomas
contents Evaluation of per-sample uncertainty quantification from neural networks is essential for decision-making involving high-risk applications. A common approach is to use the predictive distribution from Bayesian or approximation models and decompose the corresponding predictive uncertainty into epistemic (model-related) and aleatoric (data-related) components. However, additive decomposition has recently been questioned. In this work, we propose an intuitive framework for uncertainty estimation and decomposition based on the signal-to-noise ratio of class probability distributions across different model predictions. We introduce a variance-gated measure that scales predictions by a confidence factor derived from ensembles. We use this measure to discuss the existence of a collapse in the diversity of committee machines.
format Preprint
id arxiv_https___arxiv_org_abs_2509_08846
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Uncertainty Estimation using Variance-Gated Distributions
Gillis, H. Martin
Xu, Isaac
Trappenberg, Thomas
Machine Learning
Artificial Intelligence
Evaluation of per-sample uncertainty quantification from neural networks is essential for decision-making involving high-risk applications. A common approach is to use the predictive distribution from Bayesian or approximation models and decompose the corresponding predictive uncertainty into epistemic (model-related) and aleatoric (data-related) components. However, additive decomposition has recently been questioned. In this work, we propose an intuitive framework for uncertainty estimation and decomposition based on the signal-to-noise ratio of class probability distributions across different model predictions. We introduce a variance-gated measure that scales predictions by a confidence factor derived from ensembles. We use this measure to discuss the existence of a collapse in the diversity of committee machines.
title Uncertainty Estimation using Variance-Gated Distributions
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2509.08846