Statistical physics has made significant contributions to machine learning theory by providing a framework to understand the collective behaviour of large systems, such as neural networks, through probabilistic and analytical tools. The methods developed in StatPhys, particularly in the study of disordered systems, have been instrumental in analyzing learning dynamics, generalisation, and optimisation in ML models. This workshop aims to bridge these two fields by offering an accessible introduction to the key statistical and probabilistic techniques that have shaped ML theory. It will provide participants with the foundational knowledge necessary to understand the deeper, cutting-edge developments that will be explored in a subsequent event featuring leading researchers. The motivation behind this workshop is to create a clear path for students and researchers to engage with these theoretical methods, fostering a deeper understanding of the mechanisms underlying learning processes.

• 9.00 - 9.30 : Ashkan Panahi - Connecting ML and Statistics 

• 9.30 - 10.00 : Stefano Sarao Mannelli - Connecting ML and Statistical Physics

• 10.30 - 11.00 : Coffee break

• 11.00 - 12.30 : Flavio Nicoletti - Disordered systems methods in ML

• 12.30 - 14.00 : Catered lunch

• 14.00 - 15.30 : David Bosch - Gaussian methods for ML

• 15.30 - 16.00 : Coffee break

• 16.00 - 17.00 : Discussion.

Speakers

Location and Registration

The event will be held in Chalmers @ room SB-L300.

You can register here.

Note: 

We have capacity for 50 participants, and any additional registrations will be placed on a waiting list. After the registration deadline, the first 50 people who register will receive a confirmation. The rest will receive an Email indicating that they are on waiting list.