A domain-specific supercomputer for training deep neural networks

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

Google's TPU supercomputers train deep neural networks 50x faster than general-purpose supercomputers running a high-performance computing benchmark.

Related collections

Most cited references 34

Record: found
Abstract: found
Article: not found

A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play

David Silver, Thomas Hubert, Julian Schrittwieser … (2018)

The game of chess is the longest-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. By contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go by reinforcement learning from self-play. In this paper, we generalize this approach into a single AlphaZero algorithm that can achieve superhuman performance in many challenging games. Starting from random play and given no domain knowledge except the game rules, AlphaZero convincingly defeated a world champion program in the games of chess and shogi (Japanese chess), as well as Go.

0 comments Cited 363 times – based on 0 reviews      Review now

Bookmark

Record: found
Abstract: not found
Article: not found

Some methods of speeding up the convergence of iteration methods

B.T. Polyak (1964)

0 comments Cited 305 times – based on 0 reviews      Review now

Bookmark

Record: found
Abstract: not found
Article: not found

A Stochastic Approximation Method

Herbert Robbins, Sutton Monro (1951)

0 comments Cited 299 times – based on 0 reviews      Review now

Bookmark

All references

Author and article information

Journal

Title: Communications of the ACM

Abbreviated Title: Commun. ACM

Publisher: Association for Computing Machinery (ACM)

ISSN (Print): 0001-0782

ISSN (Electronic): 1557-7317

Publication date Created: June 18 2020

Publication date (Print): June 18 2020

Volume: 63

Issue: 7

Pages: 67-78

Affiliations

[1 ]Google, Mountain View, CA

[2 ]Google, Mountain View, CA and University of California, Berkeley, CA

Article

DOI: 10.1145/3360307

SO-VID: c82d3269-b631-4743-8ce6-da6d5137055c

History

Data availability:

Comments

Comment on this article

scite_

Smart Citations

Citing PublicationsSupportingMentioningContrasting

View Citations

See how this article has been cited at scite.ai

scite shows how a scientific paper has been cited by providing the context of the citation, a classification describing whether it supports, mentions, or contrasts the cited claim, and a label indicating in which section the citation was made.