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

Bibliographic Details
Main Authors: Tomaso Poggio, Qianli Liao, Andrzej Banburski
Format: Article
Language:English
Published: Nature Publishing Group 2020-02-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-020-14663-9