2024 Nobel Prize in Physics Talk
Fri. Oct. 25 12:30 PM
- Fri. Oct. 25 01:20 PM
Contact: Evan McDonough
Location: Virtual Talk, join remotely or view in-person 3M69
Speaker: Dr. Michael Toomey. Postdoctoral Fellow, MIT.
The Physics of Machine Learning & The 2024 Nobel Prize in Physics
A little-known secret in the field of artificial intelligence and machine learning is its intimate connection to physics; a bridge which is most pronounced in the context of statistical mechanics. Nowhere is this more apparent than "Boltzmann machines", pioneered by 2024 Nobel prize in physics co-recipient Geoffrey Hinton, which fall into a class of energy-based machine learning algorithms, building on earlier work by Nobel co-recipient John Hopfield. While Boltzmann machines lack state-of-the-art performance, they are useful to peer into the black box which is machine learning and understand the phenomenon of learning via the lens of physics. In this talk we will explore this connection in detail with the aim to show that tools from theoretical physics are crucial for a theoretical understanding of machine learning - and potentially vice versa.
BIO: Dr. Michael Toomey is a postdoctoral fellow at the Massachusetts Institute of Technology Center for Theoretical Physics. He completed his PhD at Brown University in 2023, including an internship at Microsoft Research, where he proposed "The Autodidactic Universe": an approach to cosmology in which the Universe learns its own physical laws. His research spans the interfaces of astrophysics, particle physics, and gravity, where he often deploys and develops tools in machine learning and artificial intelligence.
For a zoom invitation to this event, please email: an.wiebe@uwinnipeg.ca or e.mcdonough@uwinnipeg.ca.