Condensed Stein Variational Gradient Descent for Uncertainty Quantification of Neural Networks

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Padmanabha, Govinda Anantha, Safta, Cosmin, Bouklas, Nikolaos, Jones, Reese E.
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910758902693888
author Padmanabha, Govinda Anantha
Safta, Cosmin
Bouklas, Nikolaos
Jones, Reese E.
author_facet Padmanabha, Govinda Anantha
Safta, Cosmin
Bouklas, Nikolaos
Jones, Reese E.
contents We propose a Stein variational gradient descent method to concurrently sparsify, train, and provide uncertainty quantification of a complexly parameterized model such as a neural network. It employs a graph reconciliation and condensation process to reduce complexity and increase similarity in the Stein ensemble of parameterizations. Therefore, the proposed condensed Stein variational gradient (cSVGD) method provides uncertainty quantification on parameters, not just outputs. Furthermore, the parameter reduction speeds up the convergence of the Stein gradient descent as it reduces the combinatorial complexity by aligning and differentiating the sensitivity to parameters. These properties are demonstrated with an illustrative example and an application to a representation problem in solid mechanics.
format Preprint
id arxiv_https___arxiv_org_abs_2412_16462
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Condensed Stein Variational Gradient Descent for Uncertainty Quantification of Neural Networks
Padmanabha, Govinda Anantha
Safta, Cosmin
Bouklas, Nikolaos
Jones, Reese E.
Machine Learning
Computational Physics
We propose a Stein variational gradient descent method to concurrently sparsify, train, and provide uncertainty quantification of a complexly parameterized model such as a neural network. It employs a graph reconciliation and condensation process to reduce complexity and increase similarity in the Stein ensemble of parameterizations. Therefore, the proposed condensed Stein variational gradient (cSVGD) method provides uncertainty quantification on parameters, not just outputs. Furthermore, the parameter reduction speeds up the convergence of the Stein gradient descent as it reduces the combinatorial complexity by aligning and differentiating the sensitivity to parameters. These properties are demonstrated with an illustrative example and an application to a representation problem in solid mechanics.
title Condensed Stein Variational Gradient Descent for Uncertainty Quantification of Neural Networks
topic Machine Learning
Computational Physics
url https://arxiv.org/abs/2412.16462