AI Safety team at the Centre for Mathematical AI at Fields Institute for Mathematics
ActiveResearch focus
Developing rigorous mathematical foundations for AI interpretability. Research directions include mean field theories of deep learning, data attribution methods, renormalization group approaches to neural networks, learning-theoretic analysis of algorithms and geometric analysis of the learning landscape.
Open to collaboration
Yes
Looking for
Postdoctoral fellows
Join the Damaqu community to see this team's full profile.