Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)

I review unsupervised or self-supervised neural networks playing minimax games in game-theoretic settings: (i) Artificial Curiosity (AC, 1990) is based on two such networks. One network learns to generate a probability distribution over outputs, the other learns to predict effects of the outputs. Ea...

Full description

Saved in:
Bibliographic Details
Published in:Neural networks 2020-07, Vol.127, p.58-66
Main Author: Schmidhuber, Jürgen
Format: Article
Language:English
Subjects:
Artificial Curiosity
Generative Adversarial Networks
Predictability Minimization
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3
cites cdi_FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3
container_end_page 66
container_issue
container_start_page 58
container_title Neural networks
container_volume 127
creator Schmidhuber, Jürgen
description I review unsupervised or self-supervised neural networks playing minimax games in game-theoretic settings: (i) Artificial Curiosity (AC, 1990) is based on two such networks. One network learns to generate a probability distribution over outputs, the other learns to predict effects of the outputs. Each network minimizes the objective function maximized by the other. (ii) Generative Adversarial Networks (GANs, 2010-2014) are an application of AC where the effect of an output is 1 if the output is in a given set, and 0 otherwise. (iii) Predictability Minimization (PM, 1990s) models data distributions through a neural encoder that maximizes the objective function minimized by a neural predictor of the code components. I correct a previously published claim that PM is not based on a minimax game.
doi_str_mv 10.1016/j.neunet.2020.04.008
format article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2394882968</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0893608020301283</els_id><sourcerecordid>2394882968</sourcerecordid><originalsourceid>FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3</originalsourceid><addsrcrecordid>eNp9kU9vEzEQxS0EomnhGyDkY3vYZfxHG_uCFEVQKpW2BzhbXntWctisg-0NCl-iXxmnKRw5jTR6v3ma9wh5x6BlwLoPm3bCecLScuDQgmwB1AuyYGqpG75U_CVZgNKi6UDBGTnPeQMAnZLiNTkTXAgpJFuQx2ucMNkS9khXfo8p2xTsSO-w_IrpR6Y2Ic07dMelsxkzjQNdpRKG8LRbzynEHMqBXjKt4YrayVM75kjdGDOOB5pwtAU9LZE-JPTBFduH8Uh8DVPYht_VPU5POLt6Q14Nlca3z_OCfP_86dv6S3N7f32zXt02TnS8NEtpewAvmZLWKceGwTGBrO-k73Uneug0ouOOWQmKaWCCd3YJsveyxoReXJDL091dij9nzMVsQ3Y4jnbCOGfDhZZKcd2pKpUnqUsx54SD2aWwtelgGJhjFWZjTlWYYxUGpKkeFXv_7DD3W_T_oL_ZV8HHkwDrn_uAyWQXcHI1ooSuGB_D_x3-AJRFnNE</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2394882968</pqid></control><display><type>article</type><title>Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)</title><source>ScienceDirect Freedom Collection 2022-2024</source><creator>Schmidhuber, Jürgen</creator><creatorcontrib>Schmidhuber, Jürgen</creatorcontrib><description>I review unsupervised or self-supervised neural networks playing minimax games in game-theoretic settings: (i) Artificial Curiosity (AC, 1990) is based on two such networks. One network learns to generate a probability distribution over outputs, the other learns to predict effects of the outputs. Each network minimizes the objective function maximized by the other. (ii) Generative Adversarial Networks (GANs, 2010-2014) are an application of AC where the effect of an output is 1 if the output is in a given set, and 0 otherwise. (iii) Predictability Minimization (PM, 1990s) models data distributions through a neural encoder that maximizes the objective function minimized by a neural predictor of the code components. I correct a previously published claim that PM is not based on a minimax game.</description><identifier>ISSN: 0893-6080</identifier><identifier>EISSN: 1879-2782</identifier><identifier>DOI: 10.1016/j.neunet.2020.04.008</identifier><identifier>PMID: 32334341</identifier><language>eng</language><publisher>United States: Elsevier Ltd</publisher><subject>Artificial Curiosity ; Generative Adversarial Networks ; Predictability Minimization</subject><ispartof>Neural networks, 2020-07, Vol.127, p.58-66</ispartof><rights>2020 Elsevier Ltd</rights><rights>Copyright © 2020 Elsevier Ltd. All rights reserved.</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3</citedby><cites>FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3</cites><orcidid>0000-0002-1468-6758</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/32334341$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Schmidhuber, Jürgen</creatorcontrib><title>Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)</title><title>Neural networks</title><addtitle>Neural Netw</addtitle><description>I review unsupervised or self-supervised neural networks playing minimax games in game-theoretic settings: (i) Artificial Curiosity (AC, 1990) is based on two such networks. One network learns to generate a probability distribution over outputs, the other learns to predict effects of the outputs. Each network minimizes the objective function maximized by the other. (ii) Generative Adversarial Networks (GANs, 2010-2014) are an application of AC where the effect of an output is 1 if the output is in a given set, and 0 otherwise. (iii) Predictability Minimization (PM, 1990s) models data distributions through a neural encoder that maximizes the objective function minimized by a neural predictor of the code components. I correct a previously published claim that PM is not based on a minimax game.</description><subject>Artificial Curiosity</subject><subject>Generative Adversarial Networks</subject><subject>Predictability Minimization</subject><issn>0893-6080</issn><issn>1879-2782</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNp9kU9vEzEQxS0EomnhGyDkY3vYZfxHG_uCFEVQKpW2BzhbXntWctisg-0NCl-iXxmnKRw5jTR6v3ma9wh5x6BlwLoPm3bCecLScuDQgmwB1AuyYGqpG75U_CVZgNKi6UDBGTnPeQMAnZLiNTkTXAgpJFuQx2ucMNkS9khXfo8p2xTsSO-w_IrpR6Y2Ic07dMelsxkzjQNdpRKG8LRbzynEHMqBXjKt4YrayVM75kjdGDOOB5pwtAU9LZE-JPTBFduH8Uh8DVPYht_VPU5POLt6Q14Nlca3z_OCfP_86dv6S3N7f32zXt02TnS8NEtpewAvmZLWKceGwTGBrO-k73Uneug0ouOOWQmKaWCCd3YJsveyxoReXJDL091dij9nzMVsQ3Y4jnbCOGfDhZZKcd2pKpUnqUsx54SD2aWwtelgGJhjFWZjTlWYYxUGpKkeFXv_7DD3W_T_oL_ZV8HHkwDrn_uAyWQXcHI1ooSuGB_D_x3-AJRFnNE</recordid><startdate>202007</startdate><enddate>202007</enddate><creator>Schmidhuber, Jürgen</creator><general>Elsevier Ltd</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-1468-6758</orcidid></search><sort><creationdate>202007</creationdate><title>Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)</title><author>Schmidhuber, Jürgen</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Artificial Curiosity</topic><topic>Generative Adversarial Networks</topic><topic>Predictability Minimization</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Schmidhuber, Jürgen</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>MEDLINE - Academic</collection><jtitle>Neural networks</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Schmidhuber, Jürgen</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)</atitle><jtitle>Neural networks</jtitle><addtitle>Neural Netw</addtitle><date>2020-07</date><risdate>2020</risdate><volume>127</volume><spage>58</spage><epage>66</epage><pages>58-66</pages><issn>0893-6080</issn><eissn>1879-2782</eissn><abstract>I review unsupervised or self-supervised neural networks playing minimax games in game-theoretic settings: (i) Artificial Curiosity (AC, 1990) is based on two such networks. One network learns to generate a probability distribution over outputs, the other learns to predict effects of the outputs. Each network minimizes the objective function maximized by the other. (ii) Generative Adversarial Networks (GANs, 2010-2014) are an application of AC where the effect of an output is 1 if the output is in a given set, and 0 otherwise. (iii) Predictability Minimization (PM, 1990s) models data distributions through a neural encoder that maximizes the objective function minimized by a neural predictor of the code components. I correct a previously published claim that PM is not based on a minimax game.</abstract><cop>United States</cop><pub>Elsevier Ltd</pub><pmid>32334341</pmid><doi>10.1016/j.neunet.2020.04.008</doi><tpages>9</tpages><orcidid>https://orcid.org/0000-0002-1468-6758</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 0893-6080
ispartof Neural networks, 2020-07, Vol.127, p.58-66
issn 0893-6080
1879-2782
language eng
recordid cdi_proquest_miscellaneous_2394882968
source ScienceDirect Freedom Collection 2022-2024
subjects Artificial Curiosity
Generative Adversarial Networks
Predictability Minimization
title Generative Adversarial Networks are special cases of Artificial Curiosity (1990) and also closely related to Predictability Minimization (1991)
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-28T21%3A51%3A02IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Generative%20Adversarial%20Networks%20are%20special%20cases%20of%20Artificial%20Curiosity%20(1990)%20and%20also%20closely%20related%20to%20Predictability%20Minimization%20(1991)&rft.jtitle=Neural%20networks&rft.au=Schmidhuber,%20J%C3%BCrgen&rft.date=2020-07&rft.volume=127&rft.spage=58&rft.epage=66&rft.pages=58-66&rft.issn=0893-6080&rft.eissn=1879-2782&rft_id=info:doi/10.1016/j.neunet.2020.04.008&rft_dat=%3Cproquest_cross%3E2394882968%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c362t-74ab00d4184ac8c1ffc13e1b64db963b069eec2c1a4081901326a704bd4008ed3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2394882968&rft_id=info:pmid/32334341&rfr_iscdi=true