The Physics of Learning and The Learning of Physics

Abstract: AI has become all-encompassing in our society, it touches every technological aspect of our life, yet there is still a big gap in our fundamental understanding of how it can be used for the natural sciences. Under the umbrella of a recently received Nationaal Groeifonds grant for the creation of a chemical autonomas laboratories I will give an overview of several use cases, where AI algorithms can help us solve problems in the natural sciences. Specifically I will focus on material discovery and synthesis using both techniques from NeuralODE and Generative models with novel physics informed data generation models.

 

Bio: Professor Tal Kachman received his B.Sc (Chemistry), B.Sc (Physics and Mathematics), M.Sc (Engineering) from the Technion I.I.T. He obtained his Ph.D in theoretical physics from the Massachusetts Institute of technology and Technion I.I.T. Following his studies, he was a postdoctoral and then research staff member at IBM research NY working on Machine learning for healthcare and life science, fundamentals of deep learning, and quantum machine learning.

Tal was a member of the machine learning research group in AQR capital management working on alpha generation portfolio optimization and trade execution. Since 2022 Tal has been a Professor of AI in Radboud University and a visiting professor at the Yale School of Management and recently joined Optiver, a leading global market maker as a quantitative researcher . He leads the complex learning lab doing research into the fundamentals of learning theory and its application to different domains such as economics, physics, and chemistry.