Abstract: Neural networks, though called as black-box uniform approximator and difficult to interpret, have an unreasonable effectiveness in learning unknown mechanisms with bless of dimensionality, and have lots of applications. In this talk, I will introduce a recent state-of-the-art universal differential equation method that embeds neural networks into transmission models. Three applications will be shown. (1) Using deep learning techniques to estimate effective reproduction number and compared with EpiEstim and EpiNow2 method. (2) Discovering unknown human behavior change mechnisms in transmission dynamics. (3) Using deep learning techniques to solve optimal epidemic control problems by representing optimal control function as neural networks, and compared with traditional direct, indirect, and dynamic programming methods.

Pengfei Song received his Ph.D. degree in Mathematics from Xi’an Jiaotong University, China in 2020 (supervised by Prof. Yanni Xiao). During his Ph.D., he has visited the Ohio State University as a visiting scholar student for two years since September 2017 (supervised by Prof. Yuan Lou). He is now working as a Postdoctoral Fellow at York University (supervised by Prof. Jianhong Wu). Pengfei’s research interests include multi-scale modeling of complex biological phenomena; couple of deep learning and differential equations; spatial, temporal, host and network heterogeneities on the spread of infectious disease.

Personal Links:

Github: https://github.com/Song921012

Website: Pengfei's MathBio Blog (song921012.github.io) including my publications and CV.

CV: https://song921012.github.io/2021/02/01/Pengfei%20CV/