Improved uncertainty quantification for neural networks with Bayesian last layer

Simple item page
dc.contributor.authorFiedler, Felix
dc.contributor.authorLucia, Sergio
dc.date.accessioned2024-04-17T13:33:18Z
dc.date.available2024-04-17T13:33:18Z
dc.date.issued2023-11-02
dc.description.abstractUncertainty quantification is an important task in machine learning - a task in which standard neural networks (NNs) have traditionally not excelled. This can be a limitation for safety-critical applications, where uncertainty-aware methods like Gaussian processes or Bayesian linear regression are often preferred. Bayesian neural networks are an approach to address this limitation. They assume probability distributions for all parameters and yield distributed predictions. However, training and inference are typically intractable and approximations must be employed. A promising approximation is NNs with Bayesian last layer (BLL). They assume distributed weights only in the linear output layer and yield a normally distributed prediction. To approximate the intractable Bayesian neural network, point estimates of the distributed weights in all but the last layer should be obtained by maximizing the marginal likelihood. This has previously been challenging, as the marginal likelihood is expensive to evaluate in this setting. We present a reformulation of the log-marginal likelihood of a NN with BLL which allows for efficient training using backpropagation. Furthermore, we address the challenge of uncertainty quantification for extrapolation points. We provide a metric to quantify the degree of extrapolation and derive a method to improve the uncertainty quantification for these points. Our methods are derived for the multivariate case and demonstrated in a simulation study. In comparison to Bayesian linear regression with fixed features, and a Bayesian neural network trained with variational inference, our proposed method achieves the highest log-predictive density on test data.en
dc.identifier.urihttp://hdl.handle.net/2003/42444
dc.identifier.urihttp://dx.doi.org/10.17877/DE290R-24280
dc.language.isoende
dc.relation.ispartofseriesIEEE access;11
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/de
dc.subjectBayesian last layeren
dc.subjectBayesian neural networken
dc.subjectuncertainty quantificationen
dc.subject.ddc660
dc.titleImproved uncertainty quantification for neural networks with Bayesian last layeren
dc.typeTextde
dc.type.publicationtypeResearchArticlede
dcterms.accessRightsopen access
eldorado.secondarypublicationtruede
eldorado.secondarypublication.primarycitationF. Fiedler und S. Lucia, „Improved uncertainty quantification for neural networks with Bayesian last layer“, IEEE access, Bd. 11, S. 123149–123160, 2023, https://doi.org/10.1109/access.2023.3329685de
eldorado.secondarypublication.primaryidentifierhttps://doi.org/10.1109/ACCESS.2023.3329685de

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Improved_Uncertainty_Quantification_for_Neural_Networks_With_Bayesian_Last_Layer.pdf
Size:
1.86 MB
Format:
Adobe Portable Document Format
Description:
DNB
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
4.85 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Fakultät für Bio- und Chemieingenieurwesen