Next was a fantastic talk by Pierfrancesco Urbani on using dynamical mean field theory to understand feature learning and overfitting in large neural networks a
Pierfrancesco Urbani presented research on applying dynamical mean field theory to analyze feature learning and overfitting in large neural networks. The talk, held at the Harvard Center of Mathematical Sciences and Applications, aimed to bridge empirical observations with theoretical understanding. This work seeks to further clarify the mechanisms behind the impressive performance of large-scale AI models. AI
IMPACT Provides theoretical insights into the behavior of large neural networks, potentially guiding future model development.