Evaluating Adversarial Robustness with Expected Viable Performance

Evaluating Adversarial Robustness with Expected Viable Performance

We introduce a metric for evaluating the robustness of a classifier, with particular attention to adversarial perturbations, in terms of expected functionality with respect to possible adversarial perturbations. A classifier is assumed to be non-functional (that is, has a functionality of zero) with...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: McCoppin, Ryan, Dawson, Colin, Kennedy, Sean M, Blaha, Leslie M
Format: Artikel
Sprache:eng
Schlagworte:
Computer Science - Learning
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 McCoppin, Ryan
Dawson, Colin
Kennedy, Sean M
Blaha, Leslie M
description We introduce a metric for evaluating the robustness of a classifier, with particular attention to adversarial perturbations, in terms of expected functionality with respect to possible adversarial perturbations. A classifier is assumed to be non-functional (that is, has a functionality of zero) with respect to a perturbation bound if a conventional measure of performance, such as classification accuracy, is less than a minimally viable threshold when the classifier is tested on examples from that perturbation bound. Defining robustness in terms of an expected value is motivated by a domain general approach to robustness quantification.
doi_str_mv 10.48550/arxiv.2309.09928
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2309_09928</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2309_09928</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-7f111d54739b8ae4e0a36766cdae97cf5bb4588319bcecdd14e9c0c6cac634f53</originalsourceid><addsrcrecordid>eNotz8tKAzEUgOFsXEj1AVyZF5gxae7LWsYLFBQpboeT5EQD02lJpmN9e7G6-nc_fITccNZKqxS7g3LKc7sUzLXMuaW9JPfdDMMRpjx-0FWcsVQoGQb6tvfHOo1YK_3K0yftTgcME0b6nsEPSF-xpH3ZwRjwilwkGCpe_3dBtg_ddv3UbF4en9erTQPa2MYkznlU0gjnLaBEBkIbrUMEdCYk5b1U1grufMAQI5foAgs6QNBCJiUW5PZve0b0h5J3UL77X0x_xogfPDBF1g</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Evaluating Adversarial Robustness with Expected Viable Performance</title><source>arXiv.org</source><creator>McCoppin, Ryan ; Dawson, Colin ; Kennedy, Sean M ; Blaha, Leslie M</creator><creatorcontrib>McCoppin, Ryan ; Dawson, Colin ; Kennedy, Sean M ; Blaha, Leslie M</creatorcontrib><description>We introduce a metric for evaluating the robustness of a classifier, with particular attention to adversarial perturbations, in terms of expected functionality with respect to possible adversarial perturbations. A classifier is assumed to be non-functional (that is, has a functionality of zero) with respect to a perturbation bound if a conventional measure of performance, such as classification accuracy, is less than a minimally viable threshold when the classifier is tested on examples from that perturbation bound. Defining robustness in terms of an expected value is motivated by a domain general approach to robustness quantification.</description><identifier>DOI: 10.48550/arxiv.2309.09928</identifier><language>eng</language><subject>Computer Science - Learning</subject><creationdate>2023-09</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/2309.09928$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2309.09928$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>McCoppin, Ryan</creatorcontrib><creatorcontrib>Dawson, Colin</creatorcontrib><creatorcontrib>Kennedy, Sean M</creatorcontrib><creatorcontrib>Blaha, Leslie M</creatorcontrib><title>Evaluating Adversarial Robustness with Expected Viable Performance</title><description>We introduce a metric for evaluating the robustness of a classifier, with particular attention to adversarial perturbations, in terms of expected functionality with respect to possible adversarial perturbations. A classifier is assumed to be non-functional (that is, has a functionality of zero) with respect to a perturbation bound if a conventional measure of performance, such as classification accuracy, is less than a minimally viable threshold when the classifier is tested on examples from that perturbation bound. Defining robustness in terms of an expected value is motivated by a domain general approach to robustness quantification.</description><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz8tKAzEUgOFsXEj1AVyZF5gxae7LWsYLFBQpboeT5EQD02lJpmN9e7G6-nc_fITccNZKqxS7g3LKc7sUzLXMuaW9JPfdDMMRpjx-0FWcsVQoGQb6tvfHOo1YK_3K0yftTgcME0b6nsEPSF-xpH3ZwRjwilwkGCpe_3dBtg_ddv3UbF4en9erTQPa2MYkznlU0gjnLaBEBkIbrUMEdCYk5b1U1grufMAQI5foAgs6QNBCJiUW5PZve0b0h5J3UL77X0x_xogfPDBF1g</recordid><startdate>20230918</startdate><enddate>20230918</enddate><creator>McCoppin, Ryan</creator><creator>Dawson, Colin</creator><creator>Kennedy, Sean M</creator><creator>Blaha, Leslie M</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230918</creationdate><title>Evaluating Adversarial Robustness with Expected Viable Performance</title><author>McCoppin, Ryan ; Dawson, Colin ; Kennedy, Sean M ; Blaha, Leslie M</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-7f111d54739b8ae4e0a36766cdae97cf5bb4588319bcecdd14e9c0c6cac634f53</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>McCoppin, Ryan</creatorcontrib><creatorcontrib>Dawson, Colin</creatorcontrib><creatorcontrib>Kennedy, Sean M</creatorcontrib><creatorcontrib>Blaha, Leslie M</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>McCoppin, Ryan</au><au>Dawson, Colin</au><au>Kennedy, Sean M</au><au>Blaha, Leslie M</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Evaluating Adversarial Robustness with Expected Viable Performance</atitle><date>2023-09-18</date><risdate>2023</risdate><abstract>We introduce a metric for evaluating the robustness of a classifier, with particular attention to adversarial perturbations, in terms of expected functionality with respect to possible adversarial perturbations. A classifier is assumed to be non-functional (that is, has a functionality of zero) with respect to a perturbation bound if a conventional measure of performance, such as classification accuracy, is less than a minimally viable threshold when the classifier is tested on examples from that perturbation bound. Defining robustness in terms of an expected value is motivated by a domain general approach to robustness quantification.</abstract><doi>10.48550/arxiv.2309.09928</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2309.09928
ispartof
issn
language eng
recordid cdi_arxiv_primary_2309_09928
source arXiv.org
subjects Computer Science - Learning
title Evaluating Adversarial Robustness with Expected Viable Performance
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-08T13%3A55%3A34IST&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=Evaluating%20Adversarial%20Robustness%20with%20Expected%20Viable%20Performance&rft.au=McCoppin,%20Ryan&rft.date=2023-09-18&rft_id=info:doi/10.48550/arxiv.2309.09928&rft_dat=%3Carxiv_GOX%3E2309_09928%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