Random Matrices and Dynamics of Optimization in Very High Dimensions - Lecture 4
Explore random matrices and optimization dynamics in high dimensions, focusing on effective dynamics, spectral transitions, and applications in machine learning algorithms.
Towards Understanding the Implicit Regularization Effect of SGD
Explore the implicit regularization effect of Stochastic Gradient Descent, uncovering its impact on optimization and generalization in machine learning models.
Random Matrices and Dynamics of Optimization in Very High Dimensions - Lecture 2
Explorez les dynamiques d'optimisation en très haute dimension et leur application en apprentissage automatique, en mettant l'accent sur les matrices aléatoires et les statistiques résumées.
Random Matrices and Dynamics of Optimization in Very High Dimensions - Lecture 1
Explorez les algorithmes d'optimisation en haute dimension pour l'apprentissage automatique, en mettant l'accent sur la descente de gradient stochastique et les transitions spectrales des matrices aléatoires.
Gradients Look Alike: Sensitivity is Often Overestimated in DP-SGD
Explore how DP-SGD provides stronger privacy guarantees for similar datapoints in deep learning, revealing why common benchmarks leak less information than previously estimated.