Deep learning's recent history has been one of achievement: from triumphing
over humans in the game of Go to world-leading performance in image
recognition, voice recognition, translation, and other tasks. But this progress
has come with a voracious appetite for computing power. This article reports on
the computational demands of Deep Learning applications in five prominent
application areas and shows that progress in all five is strongly reliant on
increases in computing power. Extrapolating forward this reliance reveals that
progress along current lines is rapidly becoming economically, technically, and
environmentally unsustainable. Thus, continued progress in these applications
will require dramatically more computationally-efficient methods, which will
either have to come from changes to deep learning or from moving to other
machine learning methods.
%0 Generic
%1 thompson2020computational
%A Thompson, Neil C.
%A Greenewald, Kristjan
%A Lee, Keeheon
%A Manso, Gabriel F.
%D 2020
%K ai discussion dl limits
%T The Computational Limits of Deep Learning
%U http://arxiv.org/abs/2007.05558
%X Deep learning's recent history has been one of achievement: from triumphing
over humans in the game of Go to world-leading performance in image
recognition, voice recognition, translation, and other tasks. But this progress
has come with a voracious appetite for computing power. This article reports on
the computational demands of Deep Learning applications in five prominent
application areas and shows that progress in all five is strongly reliant on
increases in computing power. Extrapolating forward this reliance reveals that
progress along current lines is rapidly becoming economically, technically, and
environmentally unsustainable. Thus, continued progress in these applications
will require dramatically more computationally-efficient methods, which will
either have to come from changes to deep learning or from moving to other
machine learning methods.
@misc{thompson2020computational,
abstract = {Deep learning's recent history has been one of achievement: from triumphing
over humans in the game of Go to world-leading performance in image
recognition, voice recognition, translation, and other tasks. But this progress
has come with a voracious appetite for computing power. This article reports on
the computational demands of Deep Learning applications in five prominent
application areas and shows that progress in all five is strongly reliant on
increases in computing power. Extrapolating forward this reliance reveals that
progress along current lines is rapidly becoming economically, technically, and
environmentally unsustainable. Thus, continued progress in these applications
will require dramatically more computationally-efficient methods, which will
either have to come from changes to deep learning or from moving to other
machine learning methods.},
added-at = {2020-12-04T11:00:42.000+0100},
author = {Thompson, Neil C. and Greenewald, Kristjan and Lee, Keeheon and Manso, Gabriel F.},
biburl = {https://www.bibsonomy.org/bibtex/25f6048a20d2229806f585e86a6cb8e43/louissf},
description = {The Computational Limits of Deep Learning},
interhash = {b5918ee73d69062c886efa43797d07b7},
intrahash = {5f6048a20d2229806f585e86a6cb8e43},
keywords = {ai discussion dl limits},
note = {cite arxiv:2007.05558},
timestamp = {2020-12-04T11:00:42.000+0100},
title = {The Computational Limits of Deep Learning},
url = {http://arxiv.org/abs/2007.05558},
year = 2020
}