Theory I: Deep networks and the curse of dimensionality

TitleTheory I: Deep networks and the curse of dimensionality
Publication TypeJournal Article
Year of Publication2018
AuthorsPoggio, T, Liao, Q
JournalBulletin of the Polish Academy of Sciences: Technical Sciences
Volume66
Issue6
Keywordsconvolutional neural networks, deep and shallow networks, deep learning, function approximation
Abstract

We review recent work characterizing the classes of functions for which deep learning can be exponentially better than shallow learning. Deep convolutional networks are a special case of these conditions, though weight sharing is not the main reason for their exponential advantage.

Download:  PDF icon 02_761-774_00966_Bpast.No_.66-6_28.12.18_K1.pdf

Associated Module: 

Research Area: 

CBMM Relationship: 

  • CBMM Funded