Title: Deep Learning Optimization from First principles...
Abstract:
Deep learning optimization is often treated as alchemy, yet its success relies on fundamental mathematical structures hidden within high-dimensional landscapes. This keynote deconstructs the training process from first principles, examining the “Big Brothers” that governs optimization: Convexity (geometry), and Smoothness (dynamics).
We demonstrate how physical intuitions link geometry to performance, present the rigorous connection between Optimization and Generalization, and explain essential tools for finding robust solutions in a shifting world.
Biography :
Dr. Mohammed-Amine Koulali is a Full Professor at the National School of Applied Sciences of Oujda and an Affiliate Professor at the UM6P College of Computing. His research focuses on the application of Game Theory and Reinforcement Learning to the Internet of Vehicles. Dr. Koulali received his Ph.D. from University Mohammed V-Souissi (2012) and his M.Sc. from the University of Franche-Comté (2009). He actively serves as a reviewer and committee member for various international AI and networking conferences.