【报告题目】 When machine learning meets physics
【报 告 人】 刘军伟 教授
香港科技大学
【时 间】 2019-12-13 4:30 pm (Friday)
【地 点】 北园106报告厅
【报告摘要】 Machine learning techniques, especially the deep neural networks, have been widely used for industrial applications and also played important roles in fundamental researches. In this talk, I will explore the mutual benefits between ML and physics. I will first introduce our recently develop a new general-purpose numerical method, Self-learning Monte Carlo (SLMC), which can efficiently reduce the critical slowing down in bosonic and even completely cure it in fermionic systems. Moreover, SLMC also provides a general framework to naturally integrate the advanced ML techniques into Monte Carlo. In the second part, I will talk about how we can use physic to improve the ML techniques. I will talk our recent developed all optical neural networks, which can realize the intrinsic parallel calculations at the speed of light and are expected outperform the electronic neural networks with large system size.
【报告人简介】 Prof. Junwei Liu obtained his PhD in the department of physics, Tsinghua University, in 2014, and then he started his postdoctoral research in Massachusetts Institute of Technology. He joined Hong Kong University of Science and Technology in 2017 as an assistant professor. His research interest includes: 1) Topological materials including quantum anomalous Hall insulators, topological insulators topological crystalline insulators and topological semi-metals; 2) Quantum Monte Carlo simulations of strongly correlated systems; 3) Atomic-layer-thin ferroelectric materials and physics; 4) Two-dimensional quantum materials and phenomena; 5) Applications of machine learning in physics.