Thursday, August 18

8:00-8:30          Registration

8:30-8:45          Welcome – Abani Patra and Serge Prudhomme

8:45-9:55          Session 1, chaired by Krishna Garikipati

8:45-9:15          Erin Acquesta, Sandia National Laboratories
Adapting Verification and Validation Principles to a Credibility Process for Scientific Machine Learning

9:15-9:35          Youssef Marzouk, Massachusetts Institute of Technology
Likelihood-Free Bayesian Inference via Transportation of Measure

9:35-9:55          Michael Shields, Johns Hopkins University
Manifold Learning for High-Dimensional Uncertainty Quantification

9:55-10:15       BREAK

10:15-11:35     Session 2, chaired by Yue Yu

10:15-10:35     Dhruv Patel and Eric Darve, Stanford University
Multifidelity Monte-Carlo Sampling in Mechanics

10:35-10:55     Romit Maulik, Argonne National Laboratory
Quantifying Uncertainty in Deep Learning for Fluid Flow Reconstruction

10:55-11:15     Marta D'Elia,  Meta Reality Labs.
A Data-Driven Continuum Model for Upscaling Noisy Molecular Dynamics Displacements

11:15-11:35     Jian-xun Wang, University of Notre Dame
Predicting Parametric Spatiotemporal Dynamics by Multi-Resolution PDE Structure-Preserved Deep Learning

11:35-1:00       LUNCH (on your own)

1:00-2:10          Session 3, chaired by Abani Patra

1:00-1:30          Vipin Kumar, University of Minnesota
Knowledge-Guided Machine Learning (KGML): A New Framework for Accelerating Scientific Discovery

1:30-1:50          Varun Chandola, University at Buffalo, National Science Foundation
ChemTab: A Physics Guided Chemistry Tabulation Framework

1:50-2:10          Yue Yu, Lehigh University
Bayesian Nonlocal Operator Regression (BNOR): Towards the Characterization of Uncertainty in Heterogeneous Materials

2:10-2:20          BREAK

2:20-3:20          Session 4, chaired by Serge Prudhomme

2:20-2:40          Alex Gorodetsky, University of Michigan
Uncertainty Quantification with Multi-Fidelity Networks

2:40-3:00          Zhuotong Chen and Zheng Zhang, University of California, Santa Barbara
Self-Healing Robust Neural Networks via Closed-Loop Control

3:00-3:20          Deep Ray and Assad Oberai, University of Southern California
Conditional GANs and Their Generalizability in Physics-Based Inverse Problems

3:20-3:30          BREAK

3:30-4:30          Session 5, chaired by Jian-xun Wang

3:30-3:50          Daniele Schiavazzi, University of Notre Dame
Adaptive and Privacy Preserving Inference for Physics Based Models    

3:50-4:10          Alireza Doostan, University of Colorado Boulder
Recovering Dynamical Systems from Noisy Measurements Using Constrained Optimization

4:10-4:30          Wei Chen, Northwestern University
Physics-Informed Latent Variable Gaussian Process Modeling and Uncertainty Quantification for Adaptive Discovery and Design

4:30-5:45          Lightning Presentations, chaired by Abani Patra and Serge Prudhomme

5:45-7:15        Poster Session and Reception

Friday, August 19         

8:00-8:45          Registration                                                                                                                                                   

8:45-10:15       Session 6, chaired by Johann Guilleminot

8:45-9:15          Christian Soize, Université Gustave Eiffel
Posterior Probabilistic Learning Constrained by Stochastic PDE and Experimental Statistical Moments of Physics Observations

9:15-9:35          Tan Bui-Thanh, The University of Texas at Austin
Model-Constrained Deep Learning Methods for Forward and Inverse Problems                

9:35-9:55          Krishna Garikipati, University of Michigan
Bayesian Neural Networks for Weak Solution of PDEs with Uncertainty Quantification

9.55-10:15       Roger Ghanem, University of Southern California
  Model Refinement as Probabilistic Learning                     

10:15-10:35     BREAK

10:35-11:35     Early Career Panel, moderated by Romit Maulik and Abani Patra

As a core experience of the UQ-MLIP we convened an early career panel with a distinguished set of panelists from different federal agencies and successful recent early career participants. The basic format of the panel was short presentations from 4 successful early career participants about challenges on their way to early success followed by a panel discussion with participation from federal agency program leadership. Key participants include early career participants M. Shields, Y. Yu, A. Gorodetsky and R. Maulik. Each had a different profile of success and different challenges some deriving from their diverse background and different types of institutions. The concerns they expressed and challenges they raised were addressed by the panelists from across the federal government S. Lee (DOE), F. Fahroo (AFOSR), S. Qidwai/A. Sussman(NSF). Opportunities for mentorship, modalities for organizing and preparing a research program were all highlighted.

11:35-1:00       LUNCH (on your own)

1:00-2:20         Session 7, chaired by Serge Prudhomme

1:00-1:20          Zongren Zou and George Karniadakis, Brown University
Uncertainty Quantification in Scientific Machine Learning: Methods, Metrics, and Comparisons

1:20-1:40         Cosmin Safta, Sandia National Laboratories
Graph Convolutional Neural Networks for Modeling Materials with Microstructure        

1:40-2:00          Kathryn Maupin, Sandia National Laboratories
Informing Missing Physics with Model Form Error and Model Selection

2:00-2:20          Americo Cunha Jr., Rio de Janeiro State University
A Cross-Entropy Approximate Bayesian Computation Framework for Uncertainty Quantification on Mechanistic Epidemic Models

2:20-2:40          BREAK                                                              

2:40-4:00          Emerging Themes and Challenges Discussion, chaired by Abani Patra