Speaker: Guannan Zhang, senior staff scientist at Oak Ridge National Laboratory
Date and Time: September 27, 2:00-3:00 p.m.
Location: Coda, 9th Floor Atrium
Host: Peng Chen

Title: Generative Machine Learning Models for Uncertainty Quantification

Abstract: Generative machine learning models, including variational auto-encoders, normalizing flows, generative adversarial networks, diffusion models, have dramatically improved the quality and realism of generated content, whether it's images, text, or audio. In science and engineering, generative models can be used as powerful tools for probability density estimation or high-dimensional sampling that critical capabilities in uncertainty quantification (UQ), e.g., Bayesian inference for parameter estimation. Studies on generative models for image/audio synthesis focus on improving the quality of individual sample, which often make the generative models complicated and difficult to train. On the other hand, UQ tasks usually focus on accurate approximation of quantities of interest without worrying about the quality of any individual sample, so direct application of existing generative models to UQ tasks may lead to inaccurate approximation or unstable training process. To alleviate those challenges, we developed several new generative models for various UQ tasks, including training-free diffusion models for density estimation, and a score-based nonlinear filter for data assimilation, as well as scalable implementations of our UQ methods on OLCF’s supercomputers. We will discuss the effectiveness of those methods in various UQ tasks including density estimation for single and multi-modal distributions, learning stochastic dynamical systems, amortized inference for parameters estimation, and scalable data assimilation for atmosphere models.

Bio: Dr. Guannan Zhang is a Senior Staff Scientist in Machine Learning and Data Analytics Group at Oak Ridge National Laboratory (ORNL). He earned my Ph.D. in applied mathematics at Florida State University in 2012. He joined ORNL in 2012 as the Householder fellow in the Computer Science and Mathematics Division. He received the DOE Early Career Award in 2022. He has been holding a joint faculty appointment with the Department of Mathematics and Statistics at Auburn University since 2014, and a joint faculty appointment with Department of Mathematics at University of Tennessee since 2022. Guannan's research interests include high-dimensional approximation, uncertainty quantification, machine learning and artificial intelligence, stochastic optimization and control.