Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack

Model extraction increasingly attracts research attentions as keeping commercial AI models private can retain a competitive advantage. In some scenarios, AI models are trained proprietarily, where neither pre-trained models nor sufficient in-distribution data is publicly available. Model extraction...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Xinyi, Fang, Chengfang, Shi, Jie
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 Zhang, Xinyi
Fang, Chengfang
Shi, Jie
description Model extraction increasingly attracts research attentions as keeping commercial AI models private can retain a competitive advantage. In some scenarios, AI models are trained proprietarily, where neither pre-trained models nor sufficient in-distribution data is publicly available. Model extraction attacks against these models are typically more devastating. Therefore, in this paper, we empirically investigate the behaviors of model extraction under such scenarios. We find the effectiveness of existing techniques significantly affected by the absence of pre-trained models. In addition, the impacts of the attacker's hyperparameters, e.g. model architecture and optimizer, as well as the utilities of information retrieved from queries, are counterintuitive. We provide some insights on explaining the possible causes of these phenomena. With these observations, we formulate model extraction attacks into an adaptive framework that captures these factors with deep reinforcement learning. Experiments show that the proposed framework can be used to improve existing techniques, and show that model extraction is still possible in such strict scenarios. Our research can help system designers to construct better defense strategies based on their scenarios.
doi_str_mv 10.48550/arxiv.2104.05921
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2104_05921</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2104_05921</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-15297241146455014aac6796c44a54039354e409bac88a7342db143d4cad95f63</originalsourceid><addsrcrecordid>eNotz8FOwzAQBFBfOKDCB3DqfgAJdrx2Ym6lKqVSEVIVhDhFW9shFiVBjoHy95TCaQ4jjeYxdiF4jpVS_IriPnzmheCYc2UKcco2dRd8ewk3_ouih6GFp44SLH2C5-ED6s5Hfw31cGjdCI-983FM1LvQv8D94PwOFvsUyaYw9DBLiezrGTtpaTf68_-csPp2Uc_vsvXDcjWfrTPSpciEKkxZoBCo8XBNIJHVpdEWkRRyaaRCj9xsyVYVlRILtxUoHVpyRrVaTtj0b_aIat5jeKP43fzimiNO_gCceEfe</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack</title><source>arXiv.org</source><creator>Zhang, Xinyi ; Fang, Chengfang ; Shi, Jie</creator><creatorcontrib>Zhang, Xinyi ; Fang, Chengfang ; Shi, Jie</creatorcontrib><description>Model extraction increasingly attracts research attentions as keeping commercial AI models private can retain a competitive advantage. In some scenarios, AI models are trained proprietarily, where neither pre-trained models nor sufficient in-distribution data is publicly available. Model extraction attacks against these models are typically more devastating. Therefore, in this paper, we empirically investigate the behaviors of model extraction under such scenarios. We find the effectiveness of existing techniques significantly affected by the absence of pre-trained models. In addition, the impacts of the attacker's hyperparameters, e.g. model architecture and optimizer, as well as the utilities of information retrieved from queries, are counterintuitive. We provide some insights on explaining the possible causes of these phenomena. With these observations, we formulate model extraction attacks into an adaptive framework that captures these factors with deep reinforcement learning. Experiments show that the proposed framework can be used to improve existing techniques, and show that model extraction is still possible in such strict scenarios. Our research can help system designers to construct better defense strategies based on their scenarios.</description><identifier>DOI: 10.48550/arxiv.2104.05921</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition ; Computer Science - Cryptography and Security ; Computer Science - Learning</subject><creationdate>2021-04</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>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2104.05921$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2104.05921$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Zhang, Xinyi</creatorcontrib><creatorcontrib>Fang, Chengfang</creatorcontrib><creatorcontrib>Shi, Jie</creatorcontrib><title>Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack</title><description>Model extraction increasingly attracts research attentions as keeping commercial AI models private can retain a competitive advantage. In some scenarios, AI models are trained proprietarily, where neither pre-trained models nor sufficient in-distribution data is publicly available. Model extraction attacks against these models are typically more devastating. Therefore, in this paper, we empirically investigate the behaviors of model extraction under such scenarios. We find the effectiveness of existing techniques significantly affected by the absence of pre-trained models. In addition, the impacts of the attacker's hyperparameters, e.g. model architecture and optimizer, as well as the utilities of information retrieved from queries, are counterintuitive. We provide some insights on explaining the possible causes of these phenomena. With these observations, we formulate model extraction attacks into an adaptive framework that captures these factors with deep reinforcement learning. Experiments show that the proposed framework can be used to improve existing techniques, and show that model extraction is still possible in such strict scenarios. Our research can help system designers to construct better defense strategies based on their scenarios.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><subject>Computer Science - Cryptography and Security</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz8FOwzAQBFBfOKDCB3DqfgAJdrx2Ym6lKqVSEVIVhDhFW9shFiVBjoHy95TCaQ4jjeYxdiF4jpVS_IriPnzmheCYc2UKcco2dRd8ewk3_ouih6GFp44SLH2C5-ED6s5Hfw31cGjdCI-983FM1LvQv8D94PwOFvsUyaYw9DBLiezrGTtpaTf68_-csPp2Uc_vsvXDcjWfrTPSpciEKkxZoBCo8XBNIJHVpdEWkRRyaaRCj9xsyVYVlRILtxUoHVpyRrVaTtj0b_aIat5jeKP43fzimiNO_gCceEfe</recordid><startdate>20210412</startdate><enddate>20210412</enddate><creator>Zhang, Xinyi</creator><creator>Fang, Chengfang</creator><creator>Shi, Jie</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20210412</creationdate><title>Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack</title><author>Zhang, Xinyi ; Fang, Chengfang ; Shi, Jie</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-15297241146455014aac6796c44a54039354e409bac88a7342db143d4cad95f63</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><topic>Computer Science - Cryptography and Security</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Zhang, Xinyi</creatorcontrib><creatorcontrib>Fang, Chengfang</creatorcontrib><creatorcontrib>Shi, Jie</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Zhang, Xinyi</au><au>Fang, Chengfang</au><au>Shi, Jie</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack</atitle><date>2021-04-12</date><risdate>2021</risdate><abstract>Model extraction increasingly attracts research attentions as keeping commercial AI models private can retain a competitive advantage. In some scenarios, AI models are trained proprietarily, where neither pre-trained models nor sufficient in-distribution data is publicly available. Model extraction attacks against these models are typically more devastating. Therefore, in this paper, we empirically investigate the behaviors of model extraction under such scenarios. We find the effectiveness of existing techniques significantly affected by the absence of pre-trained models. In addition, the impacts of the attacker's hyperparameters, e.g. model architecture and optimizer, as well as the utilities of information retrieved from queries, are counterintuitive. We provide some insights on explaining the possible causes of these phenomena. With these observations, we formulate model extraction attacks into an adaptive framework that captures these factors with deep reinforcement learning. Experiments show that the proposed framework can be used to improve existing techniques, and show that model extraction is still possible in such strict scenarios. Our research can help system designers to construct better defense strategies based on their scenarios.</abstract><doi>10.48550/arxiv.2104.05921</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2104.05921
ispartof
issn
language eng
recordid cdi_arxiv_primary_2104_05921
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
Computer Science - Cryptography and Security
Computer Science - Learning
title Thief, Beware of What Get You There: Towards Understanding Model Extraction Attack
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-04T19%3A36%3A55IST&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=Thief,%20Beware%20of%20What%20Get%20You%20There:%20Towards%20Understanding%20Model%20Extraction%20Attack&rft.au=Zhang,%20Xinyi&rft.date=2021-04-12&rft_id=info:doi/10.48550/arxiv.2104.05921&rft_dat=%3Carxiv_GOX%3E2104_05921%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