Research Challenges and Opportunities at the interface of Machine Learning and Uncertainty Quantification


Workshop Objectives

The fast growth in practical applications of machine learning (ML), in a range of contexts including, e.g. physical, cyber, and socioeconomic systems, has fueled a renewed interest in ML methods over recent years. These advances and achievements depend on the availability of (1) effective algorithms, (2) significant computational capabilities and (3) large volumes of data. Despite the demonstrated successes of ML, it is well recognized that the reliability of ML decisions is often strongly impacted by data errors/noise, data gaps and partial data, modeling errors, and generalization. Accordingly, significant research efforts are targeted at probabilistic ML (PML) methods, to enable the quantification of uncertainty in trained ML model parameters and predictions. The technical challenges in PML are similar to those faced in the field of uncertainty quantification (UQ). The two fields share difficulties with both training/parameter-estimation and forward prediction, when dealing with large models with many parameters, fundamental modeling uncertainties, large data that is nonetheless often insufficiently informative, data gaps and partial/missing data, data errors and noise, dangers of overfitting, and computational costs. The purpose of this workshop is to provide a forum for exchange of information from experts in each of these two communities, exposing and discussing challenges, opportunities, and potential future research directions for advancing the state of the art across both domains.

List of Topics


Abstracts and oral presentations are not required for participation. We anticipate a limited number of oral presentations at the Workshop, with significant time devoted to breakout and discussion sessions. Participants interested in making an oral presentation are invited to submit a 2-page abstract describing their relevant research contributions. Abstracts will be reviewed, and authors notified regarding acceptance, as per the following Timeline:

The Workshop is jointly sponsored by FASTMATH and USC and is in cooperation with SIAM and USACM

Organizing Committee:

Registration Fee

The fee to attend the workshop is $100. Payment instructions will be emailed to admitted participants. Notifications together with payment information will be emailed to applicants by April 16 2018.

Register to Attend the Workshop

Abstract submission is now closed. You may still sign up to attend the workshop without submitting an abstract.

Further Information:

Last modified: Thu Apr 12 15:55:53 PDT 2018