Explore the second part of a comprehensive lecture on generalization and optimization in machine learning, presented by Misha Belkin from the University of California, San Diego. Delve into the evolving understanding of generalization in machine learning, focusing on recent developments driven by empirical findings in neural networks. Examine how these discoveries have necessitated a reevaluation of theoretical foundations. Gain insights into the optimization process using gradient descent and discover why large non-convex systems are surprisingly easy to optimize through local methods. Enhance your knowledge of key concepts in data science and artificial intelligence through this in-depth presentation, part of IPAM's Mathematics of Intelligences Tutorials at UCLA.
The Elusive Generalization and Easy Optimization - Part 2
Institute for Pure & Applied Mathematics (IPAM) via YouTube
Overview
Syllabus
Misha Belkin - The elusive generalization and easy optimization, Pt. 2 of 2 - IPAM at UCLA
Taught by
Institute for Pure & Applied Mathematics (IPAM)
Related Courses
-
The Elusive Generalization and Easy Optimization in Machine Learning - Part 1
-
The Elusive Generalization: Classical Bounds to Double Descent to Grokking
-
Distributed Optimization and Machine Learning
-
Analyzing Optimization and Generalization in Deep Learning via Trajectories of Gradient Descent
-
The Price of Implicit Bias in Robust Machine Learning
-
Memory as a Lens to Understand Efficient Learning and Optimization
