Xiangyu Chang - Provable Benefits of Overparameterization in Model Compression

preview_player
Показать описание
Contributed talk at the Workshop on the Theory of Overparameterized Machine Learning (TOPML) 2021.
Speaker: Xiangyu Chang (UC Riverside)
Talk title: Provable Benefits of Overparameterization in Model Compression: From Double Descent to Pruning Neural Networks

  1. TOPML Workshop
Рекомендации по теме
Xiangyu Chang - 0:26:43

Xiangyu Chang - Provable Benefits of Overparameterization in Model Compression

Suriya Gunasekar - 0:29:40

Suriya Gunasekar - Kernel and rich regimes in overparameterized linear models

Neural Network Compression 0:15:53

Neural Network Compression | Presented by Sultan

Surprises in High-Dimensional 1:08:19

Surprises in High-Dimensional (Overparameterized) Linear Classification

Stephan Mandt (UC 0:50:39

Stephan Mandt (UC Irvine) - Compressing Variational Bayes

Lester Mackey | 0:37:25

Lester Mackey | Advances in Distribution Compression

Compression of deep 0:33:44

Compression of deep cnn for fast and low power mobile applications

Seminar in Computer 2:05:45

Seminar in Computer Architecture - Session 6: Deep Compression & SneakySnake (Spring 2021)

Deep Learning Theory 0:28:24

Deep Learning Theory Session. Abba Kreiger

IBA: Deep Learning 1:16:09

IBA: Deep Learning for IoT - Lecture 16 : Autoencoders and Variational Autoencoders

Lecture 28: (Smoothness 1:25:11

Lecture 28: (Smoothness and Overparameterization); Learning BN; Naive Bayes;

Mikhail Belkin - 0:30:52

Mikhail Belkin - From classical bias-variance trade-off to double descent