[Moller] [Jlab-scicomp-briefs] AI Lunch Series: 12-1PM, Wednesday, February 17, 2021

[Bryan Hess]

The AI Lunch Series will be hosting Dr. João Caldeira (Google) this Wednesday for a presentation on “Deeply Uncertain: Comparing Methods of Uncertainty Quantification in Deep Learning Algorithms”. Additional information can be found at: https://www.jlab.org/AI/lunch_series/2021

12-1PM, Wednesday, February 17, 2021
Join with Bluejeans:  https://bluejeans.com/950395297

Abstract: In recent years, many methods for quantifying uncertainty in the output of deep learning algorithms have been developed. In this talk we will introduce three of the most prominent methods - Bayesian Neural Networks (BNN), Concrete Dropout (CD), and Deep Ensembles (DE) - and compare them to the standard analytic error propagation in the context of a single pendulum. We will also relate the output of these methods to terms more familiar in the context of the physical sciences. Our results highlight some pitfalls that may occur when using these UQ methods. For example, when the variation of noise in the training set is small, all methods predicted the same relative uncertainty independently of the inputs. This issue is particularly hard to avoid in BNN. On the other hand, when the test set contains samples far from the training distribution, we found that no methods sufficiently increased the uncertainties associated to their predictions. This problem was particularly clear for CD. In light of these results, we make some recommendations for usage and interpretation of UQ methods.

If you have questions, contact tennant at jlab.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/moller/attachments/20210215/978cf662/attachment-0001.html>
-------------- next part --------------
_______________________________________________
Jlab-scicomp-briefs mailing list
Jlab-scicomp-briefs at jlab.org
https://mailman.jlab.org/mailman/listinfo/jlab-scicomp-briefs