{"id":527,"date":"2020-04-01T16:58:01","date_gmt":"2020-04-01T23:58:01","guid":{"rendered":"https:\/\/dfan.engineering.asu.edu\/?page_id=527"},"modified":"2020-04-01T16:58:01","modified_gmt":"2020-04-01T23:58:01","slug":"ai-security-targeted-neural-network-attack-with-bit-trojan","status":"publish","type":"page","link":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/","title":{"rendered":"AI Security: Targeted Neural Network Attack with Bit Trojan"},"content":{"rendered":"\n<p> This repository contains a Pytorch implementation of the paper, titled \u201c&nbsp;<a href=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/CVPR2020_Trojan.pdf\">TBT: Targeted Neural Network Attack with Bit Trojan<\/a> \u201d which is published in CVPR-2019. It mainly discusses how to insert a Trojan or Back-door to a deployed DNN model in a computer through memory bit flip.<\/p>\n\n\n\n<p> <strong>[CVPR\u201920]<\/strong>&nbsp; Adnan Siraj Rakin, Zhezhi He and Deliang Fan, \u201cTBT: Targeted Neural Network Attack with Bit Trojan,\u201d&nbsp;<em>2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)<\/em>, June 16-18, 2020, Seattle, Washington, USA&nbsp;<a href=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/CVPR2020_Trojan.pdf\">[pdf]<\/a> <\/p>\n\n\n\n<p> Code is released at: <a href=\"https:\/\/github.com\/adnansirajrakin\/TBT-2020\">https:\/\/github.com\/adnansirajrakin\/TBT-2020<\/a> <\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT.png\" alt=\"\" class=\"wp-image-529\" width=\"433\" height=\"508\" srcset=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT.png 865w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT-255x300.png 255w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT-768x902.png 768w\" sizes=\"auto, (max-width: 433px) 100vw, 433px\" \/><\/figure><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/threat-model-1.png\" alt=\"\" class=\"wp-image-530\" width=\"472\" height=\"274\" srcset=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/threat-model-1.png 944w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/threat-model-1-300x174.png 300w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/threat-model-1-768x446.png 768w\" sizes=\"auto, (max-width: 472px) 100vw, 472px\" \/><\/figure><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/flow-chart.png\" alt=\"\" class=\"wp-image-531\" width=\"478\" height=\"390\" srcset=\"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/flow-chart.png 956w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/flow-chart-300x244.png 300w, https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/flow-chart-768x626.png 768w\" sizes=\"auto, (max-width: 478px) 100vw, 478px\" \/><\/figure><\/div>\n\n\n\n<h2 class=\"wp-block-heading\">Abstract<\/h2>\n\n\n\n<p>Security of modern Deep Neural Networks (DNNs) is under  severe scrutiny as the deployment of these models become  widespread in many intelligence-based applications.  Most recently, DNNs are attacked through Trojan which can effectively infect the model during the training phase and  get activated only through specific input patterns (i.e, trigger)  during inference. In this work, for the first time, we propose a novel Targeted Bit Trojan(TBT) method, which can insert a targeted neural Trojan into a DNN through bit-flip attack. Our algorithm efficiently generates a trigger specifically designed to locate certain vulnerable bits of DNN weights stored in main memory (i.e., DRAM). The objective is that once the attacker flips these vulnerable bits,  the network still operates with normal inference accuracy with benign input. However, when the attacker activates the trigger by embedding it with any input, the network is forced to classify all inputs to a certain target class. We demonstrate that flipping only several vulnerable bits identified by our method, using available bit-flip techniques (i.e, row-hammer), can transform a fully functional DNN model into a Trojan-infected model. We perform extensive experiments of CIFAR-10, SVHN and ImageNet datasets on both VGG-16 and Resnet-18 architectures. Our proposed TBT could classify 92% of test images to a target class with as little as 84 bit-flips out of 88 million weight bits on Resnet-18 for CIFAR10 dataset.<\/p>\n","protected":false},"excerpt":{"rendered":"<p class=\"mb-2\">This repository contains a Pytorch implementation of the paper, titled \u201c&nbsp;TBT: Targeted Neural Network Attack with Bit Trojan \u201d which is published in CVPR-2019. It mainly discusses how to insert a Trojan or Back-door to a deployed DNN model in a computer through memory bit flip. [CVPR\u201920]&nbsp; Adnan Siraj Rakin, Zhezhi He and Deliang Fan,&#8230;<\/p>\n","protected":false},"author":381,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"_acf_changed":false,"footnotes":""},"class_list":["post-527","page","type-page","status-publish","hentry"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.5 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan\" \/>\n<meta property=\"og:description\" content=\"This repository contains a Pytorch implementation of the paper, titled \u201c&nbsp;TBT: Targeted Neural Network Attack with Bit Trojan \u201d which is published in CVPR-2019. It mainly discusses how to insert a Trojan or Back-door to a deployed DNN model in a computer through memory bit flip. [CVPR\u201920]&nbsp; Adnan Siraj Rakin, Zhezhi He and Deliang Fan,...\" \/>\n<meta property=\"og:url\" content=\"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/\" \/>\n<meta property=\"og:site_name\" content=\"Deliang Fan\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.ece.jhu.edu\/dfan\/wp-content\/uploads\/2020\/04\/TBT.png\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"2 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/\",\"url\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/\",\"name\":\"AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/wp-content\\\/uploads\\\/sites\\\/201\\\/2020\\\/04\\\/TBT.png\",\"datePublished\":\"2020-04-01T23:58:01+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/#primaryimage\",\"url\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/wp-content\\\/uploads\\\/sites\\\/201\\\/2020\\\/04\\\/TBT.png\",\"contentUrl\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/wp-content\\\/uploads\\\/sites\\\/201\\\/2020\\\/04\\\/TBT.png\",\"width\":865,\"height\":1016},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/ai-security-targeted-neural-network-attack-with-bit-trojan\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"AI Security: Targeted Neural Network Attack with Bit Trojan\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/#website\",\"url\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/\",\"name\":\"Deliang Fan\",\"description\":\"\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/faculty.engineering.asu.edu\\\/dfan\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/","og_locale":"en_US","og_type":"article","og_title":"AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan","og_description":"This repository contains a Pytorch implementation of the paper, titled \u201c&nbsp;TBT: Targeted Neural Network Attack with Bit Trojan \u201d which is published in CVPR-2019. It mainly discusses how to insert a Trojan or Back-door to a deployed DNN model in a computer through memory bit flip. [CVPR\u201920]&nbsp; Adnan Siraj Rakin, Zhezhi He and Deliang Fan,...","og_url":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/","og_site_name":"Deliang Fan","og_image":[{"url":"https:\/\/www.ece.jhu.edu\/dfan\/wp-content\/uploads\/2020\/04\/TBT.png","type":"","width":"","height":""}],"twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/","url":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/","name":"AI Security: Targeted Neural Network Attack with Bit Trojan - Deliang Fan","isPartOf":{"@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/#website"},"primaryImageOfPage":{"@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/#primaryimage"},"image":{"@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/#primaryimage"},"thumbnailUrl":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT.png","datePublished":"2020-04-01T23:58:01+00:00","breadcrumb":{"@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/#primaryimage","url":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT.png","contentUrl":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-content\/uploads\/sites\/201\/2020\/04\/TBT.png","width":865,"height":1016},{"@type":"BreadcrumbList","@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/ai-security-targeted-neural-network-attack-with-bit-trojan\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/faculty.engineering.asu.edu\/dfan\/"},{"@type":"ListItem","position":2,"name":"AI Security: Targeted Neural Network Attack with Bit Trojan"}]},{"@type":"WebSite","@id":"https:\/\/faculty.engineering.asu.edu\/dfan\/#website","url":"https:\/\/faculty.engineering.asu.edu\/dfan\/","name":"Deliang Fan","description":"","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/faculty.engineering.asu.edu\/dfan\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"}]}},"_links":{"self":[{"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/pages\/527","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/users\/381"}],"replies":[{"embeddable":true,"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/comments?post=527"}],"version-history":[{"count":0,"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/pages\/527\/revisions"}],"wp:attachment":[{"href":"https:\/\/faculty.engineering.asu.edu\/dfan\/wp-json\/wp\/v2\/media?parent=527"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}