Complexity control by gradient descent in deep networks

Understanding the underlying mechanisms behind the successes of deep networks remains a challenge. Here, the author demonstrates an implicit regularization in training deep networks, showing that the control of complexity in the training is hidden within the optimization technique of gradient descen...

Full description

Saved in:
Bibliographic Details
Main Authors: Tomaso Poggio, Qianli Liao, Andrzej Banburski
Format: article
Language:EN
Published: Nature Portfolio 2020
Subjects:
Q
Online Access:https://doaj.org/article/a50e459c37be4efd87800b05dd1d5ce3
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Understanding the underlying mechanisms behind the successes of deep networks remains a challenge. Here, the author demonstrates an implicit regularization in training deep networks, showing that the control of complexity in the training is hidden within the optimization technique of gradient descent.