Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
This lecture delves into the analysis of generalization error in learning theory, focusing on structured data and complex architectures. The instructor explains the hidden manifold model and the use of random features in high-dimensional regression and classification. The lecture presents a detailed derivation of the generalization error expression using the replica method from statistical physics, showcasing its application to various scenarios, including the intriguing double descent behavior observed in modern machine learning.