Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry

Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this prob...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: González-Duque, Miguel, Palm, Rasmus Berg, Hauberg, Søren, Risi, Sebastian
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator González-Duque, Miguel
Palm, Rasmus Berg
Hauberg, Søren
Risi, Sebastian
description Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this problem from a geometric perspective, and provide a method for reliable interpolation and random walks in the latent spaces of Categorical VAEs based on Riemannian geometry. We test our method with "Super Mario Bros" and "The Legend of Zelda" levels, and against simpler baselines inspired by current practice. Results show that the geometry we propose is better able to interpolate and sample, reliably staying closer to parts of the latent space that decode to playable content.
doi_str_mv 10.48550/arxiv.2206.00106
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2206_00106</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2206_00106</sourcerecordid><originalsourceid>FETCH-LOGICAL-a676-6f827505fcb3186f370bbece1a16d4cd54a90ffbc5945542052bf3965eb785dd3</originalsourceid><addsrcrecordid>eNotj8tOwzAURL1hgQofwIr7AwnO4zoJOxRoQEoFEt1H147dWkrtynUR-XtKYDWj0dFIh7G7jKdljcgfKHzbrzTPuUg5z7i4ZmZDwXr4mGg-gXdAsCFnjZ_GR-i004GidTtYn52K1juaoPUuahfBOuhpaZ9HUhriPvjzbg_P1hgdLru9wJ32Bx3DfMOuDE0nffufK7Zdv2zb16R_797apz4hUYlEmDqvkKNRsshqYYqKS6mVzigTY6lGLKnhxkiFTYlY5hxzaYpGoJZVjeNYrNj93-0iOhyDPVCYh1_hYREufgDGElGF</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry</title><source>arXiv.org</source><creator>González-Duque, Miguel ; Palm, Rasmus Berg ; Hauberg, Søren ; Risi, Sebastian</creator><creatorcontrib>González-Duque, Miguel ; Palm, Rasmus Berg ; Hauberg, Søren ; Risi, Sebastian</creatorcontrib><description>Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this problem from a geometric perspective, and provide a method for reliable interpolation and random walks in the latent spaces of Categorical VAEs based on Riemannian geometry. We test our method with "Super Mario Bros" and "The Legend of Zelda" levels, and against simpler baselines inspired by current practice. Results show that the geometry we propose is better able to interpolate and sample, reliably staying closer to parts of the latent space that decode to playable content.</description><identifier>DOI: 10.48550/arxiv.2206.00106</identifier><language>eng</language><subject>Computer Science - Learning</subject><creationdate>2022-05</creationdate><rights>http://creativecommons.org/licenses/by-sa/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2206.00106$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2206.00106$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>González-Duque, Miguel</creatorcontrib><creatorcontrib>Palm, Rasmus Berg</creatorcontrib><creatorcontrib>Hauberg, Søren</creatorcontrib><creatorcontrib>Risi, Sebastian</creatorcontrib><title>Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry</title><description>Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this problem from a geometric perspective, and provide a method for reliable interpolation and random walks in the latent spaces of Categorical VAEs based on Riemannian geometry. We test our method with "Super Mario Bros" and "The Legend of Zelda" levels, and against simpler baselines inspired by current practice. Results show that the geometry we propose is better able to interpolate and sample, reliably staying closer to parts of the latent space that decode to playable content.</description><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj8tOwzAURL1hgQofwIr7AwnO4zoJOxRoQEoFEt1H147dWkrtynUR-XtKYDWj0dFIh7G7jKdljcgfKHzbrzTPuUg5z7i4ZmZDwXr4mGg-gXdAsCFnjZ_GR-i004GidTtYn52K1juaoPUuahfBOuhpaZ9HUhriPvjzbg_P1hgdLru9wJ32Bx3DfMOuDE0nffufK7Zdv2zb16R_797apz4hUYlEmDqvkKNRsshqYYqKS6mVzigTY6lGLKnhxkiFTYlY5hxzaYpGoJZVjeNYrNj93-0iOhyDPVCYh1_hYREufgDGElGF</recordid><startdate>20220531</startdate><enddate>20220531</enddate><creator>González-Duque, Miguel</creator><creator>Palm, Rasmus Berg</creator><creator>Hauberg, Søren</creator><creator>Risi, Sebastian</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20220531</creationdate><title>Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry</title><author>González-Duque, Miguel ; Palm, Rasmus Berg ; Hauberg, Søren ; Risi, Sebastian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a676-6f827505fcb3186f370bbece1a16d4cd54a90ffbc5945542052bf3965eb785dd3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>González-Duque, Miguel</creatorcontrib><creatorcontrib>Palm, Rasmus Berg</creatorcontrib><creatorcontrib>Hauberg, Søren</creatorcontrib><creatorcontrib>Risi, Sebastian</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>González-Duque, Miguel</au><au>Palm, Rasmus Berg</au><au>Hauberg, Søren</au><au>Risi, Sebastian</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry</atitle><date>2022-05-31</date><risdate>2022</risdate><abstract>Deep generative models can automatically create content of diverse types. However, there are no guarantees that such content will satisfy the criteria necessary to present it to end-users and be functional, e.g. the generated levels could be unsolvable or incoherent. In this paper we study this problem from a geometric perspective, and provide a method for reliable interpolation and random walks in the latent spaces of Categorical VAEs based on Riemannian geometry. We test our method with "Super Mario Bros" and "The Legend of Zelda" levels, and against simpler baselines inspired by current practice. Results show that the geometry we propose is better able to interpolate and sample, reliably staying closer to parts of the latent space that decode to playable content.</abstract><doi>10.48550/arxiv.2206.00106</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2206.00106
ispartof
issn
language eng
recordid cdi_arxiv_primary_2206_00106
source arXiv.org
subjects Computer Science - Learning
title Mario Plays on a Manifold: Generating Functional Content in Latent Space through Differential Geometry
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T02%3A54%3A14IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Mario%20Plays%20on%20a%20Manifold:%20Generating%20Functional%20Content%20in%20Latent%20Space%20through%20Differential%20Geometry&rft.au=Gonz%C3%A1lez-Duque,%20Miguel&rft.date=2022-05-31&rft_id=info:doi/10.48550/arxiv.2206.00106&rft_dat=%3Carxiv_GOX%3E2206_00106%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true