In this paper we introduce new bounds on the approximation of functions in
deep networks and in doing so introduce some new deep network architectures for
function approximation. These results give some theoretical insight into the
success of autoencoders and ResNets.
%0 Generic
%1 mccane2018approximation
%A McCane, Brendan
%A Szymanski, Lech
%D 2018
%K theory
%T Some Approximation Bounds for Deep Networks
%U http://arxiv.org/abs/1803.02956
%X In this paper we introduce new bounds on the approximation of functions in
deep networks and in doing so introduce some new deep network architectures for
function approximation. These results give some theoretical insight into the
success of autoencoders and ResNets.
@misc{mccane2018approximation,
abstract = {In this paper we introduce new bounds on the approximation of functions in
deep networks and in doing so introduce some new deep network architectures for
function approximation. These results give some theoretical insight into the
success of autoencoders and ResNets.},
added-at = {2018-03-12T09:57:06.000+0100},
author = {McCane, Brendan and Szymanski, Lech},
biburl = {https://www.bibsonomy.org/bibtex/268271e5bd104cea775acbd6d9aed8f1d/jk_itwm},
description = {Some Approximation Bounds for Deep Networks},
interhash = {14e38d19acda2a4fdabead02cdbf1d30},
intrahash = {68271e5bd104cea775acbd6d9aed8f1d},
keywords = {theory},
note = {cite arxiv:1803.02956Comment: 9 pages},
timestamp = {2018-03-12T09:57:06.000+0100},
title = {Some Approximation Bounds for Deep Networks},
url = {http://arxiv.org/abs/1803.02956},
year = 2018
}