Past Event: Babuška Forum

Physics Informed Neural Architectures

Eldad Haber, University of British Columbia

Abstract

Neural networks are considered the main workhorse for many 

machine learning algorithms with applications ranging from computer vision to social media.

Architectures for such networks vary significantly, and in many cases, without much theoretical grounds. 

In this talk we show that different architectures can be motivated and explained by physical analogs and dynamical systems, which allows us to explore new architectures that are able to deal with new problems that traditional networks are having difficulties to solve.

Biography

Eldad Haber is a scientific an NSERC Industrial Research Chair at the University of British Columbia. Eldad is working in the field of computational inverse problems with applications in machine learning, geosciences and medical imaging. Over the last 20 years, Eldad has written various commercial software packages that have been widely adopted by industry. Eldad has written or co-authored over 150 peer reviewed publications on computational problems and is a U.S. Department of Energy Career Award recipient. After completing his Ph.D, he spent several years as a research scientist with Schlumberger and nine years at Emory University in Atlanta at the Department of Mathematics and Computer Science. In 2011, Eldad co-founded Computational Geosciences Inc and in 2017 he co-founded Xtract.ai. In 2022 Eldad was awarded a SIAM Fellow on his work on computational inverse problems.