Exact Asymptotics and Universality for Gradient Flows and Empirical Risk Minimizers

Channel:

Subscribers:

68,700

Published on December 8, 2021 7:14:02 AM ● Video Link: https://www.youtube.com/watch?v=6D7ePwgy68g

Duration: 54:20

1,276 views

Andrea Montanari (Stanford University)
https://simons.berkeley.edu/talks/tba-152
Deep Learning Theory Symposium

Other Videos By Simons Institute for the Theory of Computing

2021-12-15	Lightning Talks by Research Fellows
2021-12-15	Welcome
2021-12-15	Lightning Talks by Industry Partners
2021-12-11	Foundations for a Democratic Metaverse
2021-12-09	The Thresholding Scheme for Mean Curvature Flow and de Giorgi’s Ideas for Gradient Flows
2021-12-08	Representation Costs of Linear Neural Networks: Analysis and Design
2021-12-08	On the Cryptographic Hardness of Learning One-Hidden Layer Neural Networks
2021-12-08	Is Overfitting Actually Benign? On the Consistency of Interpolating Methods
2021-12-08	Universality of Neural Networks
2021-12-08	The Spectrum of Nonlinear Random Matrices for Ultra-Wide Neural Networks
2021-12-08	Exact Asymptotics and Universality for Gradient Flows and Empirical Risk Minimizers
2021-12-07	Learning Staircases
2021-12-07	Sharp Matrix Concentration
2021-12-07	Non-Parametric Convergence Rates for Plain Vanilla Stochastic Gradient Descent
2021-12-07	Self-Training Converts Weak Learners to Strong Learners in Mixture Models
2021-12-07	Graphon Neural Networks and the Transferability of Graph Neural Networks
2021-12-07	Adaptive Wavelet Distillation from DNNs and Dictionary Learning
2021-12-06	Uniform Convergence of Interpolators: Gaussian Width, Norm Bounds, and Benign Overfitting
2021-12-06	Foundations of Computing: Scott Aaronson \| Polylogues
2021-12-04	Equivariant Machine Learning Structured Like Classical Physics
2021-12-03	Equivariant RL

Tags:

Simons Institute

theoretical computer science

UC Berkeley

Computer Science

Theory of Computation

Theory of Computing

Deep Learning Theory Symposium

Andrea Montanari