Deep learning models have developed many remarkable achievements in recent years. In this talk, we will discuss one type of deep learning model, the normalizing flow models, and its applications in physics. Flow models are parameterized invertible transformations with a traceable probability distribution. Thus, they can be learned to perform certain transformations to simplify the targeted problem. We will see that a learned RG transformation can speed up Monte Carlo sampling, and one can use learned symplectic transformations to automate MD analysis.
Dr. Shuo-Hui Li received his PhD in Theoretical Physics in 2020 from the Institute of Physics, Chinese Academy of Sciences, under the supervision of Prof. Lei WANG. He joined HKUST as a Postdoctoral Fellow in 2021. His works are mainly in the joint areas of computational physics and machine learning. His current research focuses on quantum machine learning and machine learning in physics.
To request for meeting link, please write to firstname.lastname@example.org.