AM Seminar: Uncertainty Quantification and Machine Learning of Physical Laws Hidden Behind Noisy Data

Speaker Name: 
Guang Lin
Speaker Title: 
Associate Professor
Speaker Organization: 
Purdue University
Start Time: 
Monday, April 29, 2019 - 4:00pm
End Time: 
Monday, April 29, 2019 - 5:00pm
Location: 
BE 372
Organizer: 
Daniele Venturi

Abstract In this talk, I will present a new data-driven paradigm on how to quantify structural uncertainty (model-form uncertainty) and learn physical laws hidden behind noisy data in complex systems governed by partial differential equations (PDEs). The key idea is to identity and approximate the important features of the PDEs using Bayesian machine learning algorithms. In particular, I will discuss Bayesian sparse feature selection and parameter estimation in the presence of noisy data. Numerical experiments demonstrating the robustness of the learning algorithms with respect to noise and data size, and their ability to learn PDEs models are presented and discussed. 

Short Bio Prof. Lin is the director of Purdue data science consulting service, associate professor of mathematics and mechanical engineering at Purdue University, with a courtesy appointment in Statistics. He earned his B.S. in mechanics from Zhejiang University in 1997, and a PhD in applied mathematics from Brown University in 2007. In 2016 he was the recipient of the NSF faculty early career development award in recognition of his work on uncertainty quantifation and big data. Lin also received the mathematical biosciences institute early career award (2015), the DOE Ronald L. Brodzinsky award (2012), the ASCR leadership computing challenge award (2010). He hase served as associate editor of the SIAM journal on multiscale modeling and simulation and in the editorial board of many international journals.