Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning

This paper investigates the use of probabilistic neural networks (PNNs) to model aleatoric uncertainty, which refers to the inherent variability in the input-output relationships of a system, often characterized by unequal variance or heteroscedasticity. Unlike traditional neural networks that produ...

Full description

Saved in:
Bibliographic Details
Main Authors: Farhad Pourkamali-Anaraki, Jamal F. Husseini, Scott E. Stapleton
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10771767/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850126956274647040
author Farhad Pourkamali-Anaraki
Jamal F. Husseini
Scott E. Stapleton
author_facet Farhad Pourkamali-Anaraki
Jamal F. Husseini
Scott E. Stapleton
author_sort Farhad Pourkamali-Anaraki
collection DOAJ
description This paper investigates the use of probabilistic neural networks (PNNs) to model aleatoric uncertainty, which refers to the inherent variability in the input-output relationships of a system, often characterized by unequal variance or heteroscedasticity. Unlike traditional neural networks that produce deterministic outputs, PNNs generate probability distributions for the target variable, allowing the determination of both predicted means and variances in regression scenarios. Contributions of this paper include the development of a probabilistic distance metric to optimize PNN architecture, and the deployment of PNNs in controlled data sets as well as a practical material science case involving fiber-reinforced composites. The findings confirm that PNNs effectively model aleatoric uncertainty, proving to be more appropriate than the commonly employed Gaussian process regression for this purpose. Specifically, in a real-world scientific machine learning context, PNNs yield remarkably accurate output mean estimates with R-squared scores approaching 0.97, and their predicted variances exhibit a high correlation coefficient of nearly 0.77, closely matching observed data variances. Hence, this research contributes to the ongoing exploration of leveraging the sophisticated representational capacity of neural networks to delineate complex input-output relationships in scientific problems.
format Article
id doaj-art-0bacf2bb71724e739fe8fb0c7c7a34d8
institution OA Journals
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-0bacf2bb71724e739fe8fb0c7c7a34d82025-08-20T02:33:48ZengIEEEIEEE Access2169-35362024-01-011217881617883110.1109/ACCESS.2024.350873510771767Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine LearningFarhad Pourkamali-Anaraki0https://orcid.org/0000-0003-4078-1676Jamal F. Husseini1Scott E. Stapleton2Department of Mathematical and Statistical Sciences, University of Colorado at Denver, Denver, CO, USADepartment of Mechanical and Industrial Engineering, University of Massachusetts at Lowell, Lowell, MA, USADepartment of Mechanical and Industrial Engineering, University of Massachusetts at Lowell, Lowell, MA, USAThis paper investigates the use of probabilistic neural networks (PNNs) to model aleatoric uncertainty, which refers to the inherent variability in the input-output relationships of a system, often characterized by unequal variance or heteroscedasticity. Unlike traditional neural networks that produce deterministic outputs, PNNs generate probability distributions for the target variable, allowing the determination of both predicted means and variances in regression scenarios. Contributions of this paper include the development of a probabilistic distance metric to optimize PNN architecture, and the deployment of PNNs in controlled data sets as well as a practical material science case involving fiber-reinforced composites. The findings confirm that PNNs effectively model aleatoric uncertainty, proving to be more appropriate than the commonly employed Gaussian process regression for this purpose. Specifically, in a real-world scientific machine learning context, PNNs yield remarkably accurate output mean estimates with R-squared scores approaching 0.97, and their predicted variances exhibit a high correlation coefficient of nearly 0.77, closely matching observed data variances. Hence, this research contributes to the ongoing exploration of leveraging the sophisticated representational capacity of neural networks to delineate complex input-output relationships in scientific problems.https://ieeexplore.ieee.org/document/10771767/Probabilistic neural networksaleatoric uncertaintyprediction intervalnetwork architecture optimization
spellingShingle Farhad Pourkamali-Anaraki
Jamal F. Husseini
Scott E. Stapleton
Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
IEEE Access
Probabilistic neural networks
aleatoric uncertainty
prediction interval
network architecture optimization
title Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
title_full Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
title_fullStr Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
title_full_unstemmed Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
title_short Probabilistic Neural Networks (PNNs) for Modeling Aleatoric Uncertainty in Scientific Machine Learning
title_sort probabilistic neural networks pnns for modeling aleatoric uncertainty in scientific machine learning
topic Probabilistic neural networks
aleatoric uncertainty
prediction interval
network architecture optimization
url https://ieeexplore.ieee.org/document/10771767/
work_keys_str_mv AT farhadpourkamalianaraki probabilisticneuralnetworkspnnsformodelingaleatoricuncertaintyinscientificmachinelearning
AT jamalfhusseini probabilisticneuralnetworkspnnsformodelingaleatoricuncertaintyinscientificmachinelearning
AT scottestapleton probabilisticneuralnetworkspnnsformodelingaleatoricuncertaintyinscientificmachinelearning