学术讲座
报告题目：Personal voyage of AI4Science: from predicting nonequilibrium dynamic phase transitions to deciphering bacteriophage genomes
报告时间：2025-03-14 10:00
报告人： 闫嘉伟 博士
报告地点：卢嘉锡楼202报告厅
转播地点：翔安校区能源材料大楼3号楼会议室1、会议室6，漳州校区生化主楼307教室
报告摘要：
Artificial intelligence is being increasingly integrated into scientific discovery to augment and accelerate research. In this talk I will focus on two distinct approaches of applying machine learning techniques to scientific discoveries. The first part will focus on computationally sampling the collective, dynamical fluctuations that lead to nonequilibrium pattern formation. Such sampling requires probing rare regions of trajectory space which becomes intractable as the system size grows. We propose a machine learning algorithm that samples rare trajectories and estimates the associated large deviation functions using a many-body control force by leveraging the flexible function representation provided by deep neural networks, importance sampling in trajectory space, and stochastic optimal control theory. We show that this approach scales to many hundreds of interacting particles and remains robust at dynamical phase transitions. In the second part, I will talk about our recent work on using language models to decipher and generate bacteriophage genomes. Our multiscale transformer model, is pre-trained on unannotated bacteriophage genomes with nucleotide-level tokenisation. We demonstrate the foundational capabilities of our model including the prediction of essential genes, genetic variant effects, regulatory element activity, and taxonomy of unannotated sequences.
报告人简介：
YAN Jiawei is a biomedical AI researcher in a biotech company in Shanghai. He works generally on applying machine learning techniques to physical and biological problems. He graduated from the School of Life Sciences at Peking University in 2015. He earned his PhD in 2020 at Harvard with Prof. Johan Paulsson, where he derived a mathematical inequality bounding the pattern of fluctuations that any generic biochemical reaction networks can arise. Before returning to China in 2022, he finished his postdoc training at the Department of Chemistry, Stanford, with Prof. Grant M. Rotskoff, where he developed machine learning algorithm for sampling rare events in nonequilibrium physical systems near dynamic phase transitions.
欢迎老师同学们积极参加!
化学化工学院
