SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective

Researchers have proposed various methods for visually interpreting the Convolutional Neural Network (CNN) via saliency maps, which include Class-Activation-Map (CAM) based approaches as a leading family. However, in terms of the internal design logic, existing CAM-based approaches often overlook th...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xuan, Xiwei, Deng, Ziquan, Lin, Hsuan-Tien, Kong, Zhaodan, Ma, Kwan-Liu
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 Xuan, Xiwei
Deng, Ziquan
Lin, Hsuan-Tien
Kong, Zhaodan
Ma, Kwan-Liu
description Researchers have proposed various methods for visually interpreting the Convolutional Neural Network (CNN) via saliency maps, which include Class-Activation-Map (CAM) based approaches as a leading family. However, in terms of the internal design logic, existing CAM-based approaches often overlook the causal perspective that answers the core "why" question to help humans understand the explanation. Additionally, current CNN explanations lack the consideration of both necessity and sufficiency, two complementary sides of a desirable explanation. This paper presents a causality-driven framework, SUNY, designed to rationalize the explanations toward better human understanding. Using the CNN model's input features or internal filters as hypothetical causes, SUNY generates explanations by bi-directional quantifications on both the necessary and sufficient perspectives. Extensive evaluations justify that SUNY not only produces more informative and convincing explanations from the angles of necessity and sufficiency, but also achieves performances competitive to other approaches across different CNN architectures over large-scale datasets, including ILSVRC2012 and CUB-200-2011.
doi_str_mv 10.48550/arxiv.2303.00244
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2303_00244</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2303_00244</sourcerecordid><originalsourceid>FETCH-LOGICAL-a674-925285e07daa6ba621e45c43c24905ed3458b975d83ca57256ded0052175e6e53</originalsourceid><addsrcrecordid>eNotj09Lw0AUxHPxINUP4Mn9Aomb3X35460Eq4VShVbBU3jdvMBikg27SdRvbxp7GoYZhvkFwV3MI5UB8Ad0P2aKhOQy4lwodR2Mh_f95yNbsw_jR2zYthvI9Y4GHIzt2MZhS9_WfbHaOlbYbrLNeE7m6p5Gt8hwLnhWO9synL0m79H9Muwqdhjr2mhD3cDeyPme9GAmugmuamw83V50FRw3T8fiJdy9Pm-L9S7EJFVhLkBkQDytEJMTJiImBVpJLVTOgSqpIDvlKVSZ1AipgKSiinMQcQqUEMhVcP8_u3CXvTPt_Ks885cLv_wDrn9WsQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective</title><source>arXiv.org</source><creator>Xuan, Xiwei ; Deng, Ziquan ; Lin, Hsuan-Tien ; Kong, Zhaodan ; Ma, Kwan-Liu</creator><creatorcontrib>Xuan, Xiwei ; Deng, Ziquan ; Lin, Hsuan-Tien ; Kong, Zhaodan ; Ma, Kwan-Liu</creatorcontrib><description>Researchers have proposed various methods for visually interpreting the Convolutional Neural Network (CNN) via saliency maps, which include Class-Activation-Map (CAM) based approaches as a leading family. However, in terms of the internal design logic, existing CAM-based approaches often overlook the causal perspective that answers the core "why" question to help humans understand the explanation. Additionally, current CNN explanations lack the consideration of both necessity and sufficiency, two complementary sides of a desirable explanation. This paper presents a causality-driven framework, SUNY, designed to rationalize the explanations toward better human understanding. Using the CNN model's input features or internal filters as hypothetical causes, SUNY generates explanations by bi-directional quantifications on both the necessary and sufficient perspectives. Extensive evaluations justify that SUNY not only produces more informative and convincing explanations from the angles of necessity and sufficiency, but also achieves performances competitive to other approaches across different CNN architectures over large-scale datasets, including ILSVRC2012 and CUB-200-2011.</description><identifier>DOI: 10.48550/arxiv.2303.00244</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2023-03</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2303.00244$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2303.00244$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Xuan, Xiwei</creatorcontrib><creatorcontrib>Deng, Ziquan</creatorcontrib><creatorcontrib>Lin, Hsuan-Tien</creatorcontrib><creatorcontrib>Kong, Zhaodan</creatorcontrib><creatorcontrib>Ma, Kwan-Liu</creatorcontrib><title>SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective</title><description>Researchers have proposed various methods for visually interpreting the Convolutional Neural Network (CNN) via saliency maps, which include Class-Activation-Map (CAM) based approaches as a leading family. However, in terms of the internal design logic, existing CAM-based approaches often overlook the causal perspective that answers the core "why" question to help humans understand the explanation. Additionally, current CNN explanations lack the consideration of both necessity and sufficiency, two complementary sides of a desirable explanation. This paper presents a causality-driven framework, SUNY, designed to rationalize the explanations toward better human understanding. Using the CNN model's input features or internal filters as hypothetical causes, SUNY generates explanations by bi-directional quantifications on both the necessary and sufficient perspectives. Extensive evaluations justify that SUNY not only produces more informative and convincing explanations from the angles of necessity and sufficiency, but also achieves performances competitive to other approaches across different CNN architectures over large-scale datasets, including ILSVRC2012 and CUB-200-2011.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj09Lw0AUxHPxINUP4Mn9Aomb3X35460Eq4VShVbBU3jdvMBikg27SdRvbxp7GoYZhvkFwV3MI5UB8Ad0P2aKhOQy4lwodR2Mh_f95yNbsw_jR2zYthvI9Y4GHIzt2MZhS9_WfbHaOlbYbrLNeE7m6p5Gt8hwLnhWO9synL0m79H9Muwqdhjr2mhD3cDeyPme9GAmugmuamw83V50FRw3T8fiJdy9Pm-L9S7EJFVhLkBkQDytEJMTJiImBVpJLVTOgSqpIDvlKVSZ1AipgKSiinMQcQqUEMhVcP8_u3CXvTPt_Ks885cLv_wDrn9WsQ</recordid><startdate>20230301</startdate><enddate>20230301</enddate><creator>Xuan, Xiwei</creator><creator>Deng, Ziquan</creator><creator>Lin, Hsuan-Tien</creator><creator>Kong, Zhaodan</creator><creator>Ma, Kwan-Liu</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230301</creationdate><title>SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective</title><author>Xuan, Xiwei ; Deng, Ziquan ; Lin, Hsuan-Tien ; Kong, Zhaodan ; Ma, Kwan-Liu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a674-925285e07daa6ba621e45c43c24905ed3458b975d83ca57256ded0052175e6e53</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Xuan, Xiwei</creatorcontrib><creatorcontrib>Deng, Ziquan</creatorcontrib><creatorcontrib>Lin, Hsuan-Tien</creatorcontrib><creatorcontrib>Kong, Zhaodan</creatorcontrib><creatorcontrib>Ma, Kwan-Liu</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Xuan, Xiwei</au><au>Deng, Ziquan</au><au>Lin, Hsuan-Tien</au><au>Kong, Zhaodan</au><au>Ma, Kwan-Liu</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective</atitle><date>2023-03-01</date><risdate>2023</risdate><abstract>Researchers have proposed various methods for visually interpreting the Convolutional Neural Network (CNN) via saliency maps, which include Class-Activation-Map (CAM) based approaches as a leading family. However, in terms of the internal design logic, existing CAM-based approaches often overlook the causal perspective that answers the core "why" question to help humans understand the explanation. Additionally, current CNN explanations lack the consideration of both necessity and sufficiency, two complementary sides of a desirable explanation. This paper presents a causality-driven framework, SUNY, designed to rationalize the explanations toward better human understanding. Using the CNN model's input features or internal filters as hypothetical causes, SUNY generates explanations by bi-directional quantifications on both the necessary and sufficient perspectives. Extensive evaluations justify that SUNY not only produces more informative and convincing explanations from the angles of necessity and sufficiency, but also achieves performances competitive to other approaches across different CNN architectures over large-scale datasets, including ILSVRC2012 and CUB-200-2011.</abstract><doi>10.48550/arxiv.2303.00244</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2303.00244
ispartof
issn
language eng
recordid cdi_arxiv_primary_2303_00244
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Computer Vision and Pattern Recognition
title SUNY: A Visual Interpretation Framework for Convolutional Neural Networks from a Necessary and Sufficient Perspective
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-01T13%3A06%3A29IST&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=SUNY:%20A%20Visual%20Interpretation%20Framework%20for%20Convolutional%20Neural%20Networks%20from%20a%20Necessary%20and%20Sufficient%20Perspective&rft.au=Xuan,%20Xiwei&rft.date=2023-03-01&rft_id=info:doi/10.48550/arxiv.2303.00244&rft_dat=%3Carxiv_GOX%3E2303_00244%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