[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-119580-en":3,"doc-seo-119580-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":26,"seo_description":14,"update_tm":27,"read_time":28},119580,687197207919,"Theodora","https://ap-avatar.wpscdn.com/avatar/a000253d6f5f7c60be?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779446848396160552",6,"Technology","AdvSecureNet - A Python Toolkit for Adversarial Machine Learning","Machine learning models remain exposed to adversarial attacks created by subtly modifying inputs to mislead predictions. AdvSecureNet delivers a PyTorch-based, research-focused toolkit designed to strengthen experimentation with features that prior tools often miss. It natively supports multi-GPU execution, provides both CLI and API interfaces, and uses external YAML configuration for flexible, reproducible workflows. The toolkit includes multiple attack and defense methods plus evaluation metrics, supported by disciplined software engineering practices. ","AdvSecureNet: A Python Toolkit for Adversarial Machine Learning  \narXiv :2409 .02629v 1 [ cs .CV] 4 Sep 2024  \nAdvSecureNet: A Python Toolkit for Adversarial Machine  \nLearning  \nMelih Catal [melihcatal@gmail.com](melihcatal@gmail.com)  \nSoftware Evolution and Architecture Lab University of Zurich, Switzerland  \nManuel G¨unther [guenther@ifi.uzh.ch](guenther@ifi.uzh.ch)  \nArti􀀌cial Intelligence and Machine Learning Group University of Zurich, Switzerland  \nAbstract  \nMachine learning models are vulnerable to adversarial attacks. Several tools have been developed to research these vulnerabilities, but they often lack comprehensive features and flexibility. We introduce AdvSecureNet, a PyTorch based toolkit for adversarial machine learning that is the first to natively support multi-GPU setups for attacks, defenses, and evaluation. It is the first toolkit that supports both CLI and API interfaces and external YAML configuration files to enhance versatility and reproducibility. The toolkit includes multiple attacks, defenses and evaluation metrics. Rigiorous software engineering practices are followed to ensure high code quality and maintainability. The project is available as an open-source project on GitHub at [https://github.com/melihcatal/advsecurenet](https://github.com/melihcatal/advsecurenet) and installable via PyPI.  \nKeywords: Adversarial Machine Learning, Trustworthy AI, Research Toolkit, PyTorch  \n1 Introduction  \nMachine learning models are widely used in fields such as self-driving cars (Bojarski et al. , 2016), facial recognition (Parmar and Mehta, 2013; G¨unther et al., 2016), and medical imaging (Mintz and Brodie, 2019), as well as in natural language processing tasks like chatbots (Brown et al., 2020) and translation services (Popel et al., 2020) . However, these models are vulnerable to adversarial attacks – subtle input modifications that can deceive the models (Goodfellow et al., 2015; Szegedy et al., 2013), which can compromise their integrity, confidentiality, or availability (Khalid et al., 2020) .  \nSeveral libraries, such as ART (Nicolae et al., 2018), AdverTorch (Ding et al., 2019), and CleverHans (Papernot et al., 2018), have been developed to research these vulnerabilities by providing tools for implementing attacks, defenses, and evaluation metrics. However, these libraries often lack key features necessary for comprehensive research and experimentation, such as native multi-GPU support, integrated CLI and API interfaces, and support for external configuration files.  \nTo address these limitations, we introduce AdvSecureNet (Adversarial Secure Networks), a comprehensive and flexible Python toolkit that supports multiple adversarial attacks, defenses, and evaluation metrics, optimized for multi-GPU setups. It includes a command-line interface (CLI) and an application programming interface (API), providing users with versatile options for experimentation and research. This paper outlines the  \nCatal and G¨unther  \n\n| Feature AdvSecureNet IBM Art AdverTorch SecML FoolBox Ares | CleverHans |\n| --- | --- |\n| Actively Maintained X X 􀀂 􀀂 􀀂 􀀂 | 􀀂 |\n| Last Year of Contribution 2024 2024 2022 2024 2024 2023 | 2023 |\n| Pytorch Support X X X X X X | X |\n| Tensor􀀍ow Support 􀀂 X 􀀂 X X 􀀂 | X |\n| Number of Adversarial Attacks 8 60 17 391 31 28 | 8 |\n| Number of Defenses 2 37 3 - - 3 | 1 |\n| Number of Evaluation Metrics 6 5 - - 2 1 | 2 |\n| Integrated Multi-GPU Support X 􀀂 􀀂 􀀂 􀀂 Limited2 | 􀀂 |\n| API Usage X X X X X X | X |\n| CLI Usage X 􀀂 􀀂 􀀂 􀀂 Limited2 | 􀀂 |\n| External Con􀀌g File X 􀀂 􀀂 􀀂 􀀂 Limited2 | 􀀂 |\n| GH Stars 2 4.6k 1.3k 138 2.7k 468 | 6.1k |\n| GH Forks 0 1. 1k 193 23 422 88 | 1.4k |\n| Number of Contributors 1 105 17 8 32 6 | 110 |\n| Number of Citations - 571 222 14 677 291 | 400 |\n\nTable 1: Feature Comparison of AdvSecureNet vs. Existing Libraries (26.06.2024)  \nfeatures, design, and contributions of AdvSecureNet to the adversarial machine learning community.  \n2 AdvSecureNet Features  \nAdversarial Attacks and Defenses: ","cbCaicYYzeccW4Wv","https://ap.wps.com/l/cbCaicYYzeccW4Wv","pdf",109432,1,"English","en",105,"# Introduction\n## AdvSecureNet Features\n### Adversarial Attacks and Defenses\n### Evaluation Metrics\n### Multi-GPU Support\n### Interfaces and Configuration\n### Built-in Models, Datasets and Target Generation\n# Feature Comparison","[{\"question\":\"What problem does AdvSecureNet address in adversarial machine learning?\",\"answer\":\"It targets the gap in existing libraries that lack flexible, comprehensive features for adversarial attack, defense, and evaluation research. AdvSecureNet focuses on usability, extensibility, and multi-GPU capability.\"},{\"question\":\"What interfaces and configuration options does AdvSecureNet provide?\",\"answer\":\"AdvSecureNet supports both command-line interface (CLI) and application programming interface (API). It also enables external YAML configuration files to improve versatility and reproducibility.\"},{\"question\":\"How does AdvSecureNet evaluate adversarial robustness and attack effectiveness?\",\"answer\":\"It provides evaluation metrics including accuracy, robustness, transferability, and similarity. Robustness measures resistance to attacks, while transferability assesses deception across different models and similarity uses perceptual measures such as PSNR and SSIM.\"}]","AdvSecureNet - A Python Toolkit for Adversarial Machine Learning | PDF",1785725095,15,{"code":4,"msg":30,"data":31},"ok",{"site_id":23,"language":22,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"advsecurenet-a-python-toolkit-for-adversarial-machine-learning","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/advsecurenet-a-python-toolkit-for-adversarial-machine-learning/119580/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":22,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does AdvSecureNet address in adversarial machine learning?","Question",{"text":74,"@type":75},"It targets the gap in existing libraries that lack flexible, comprehensive features for adversarial attack, defense, and evaluation research. AdvSecureNet focuses on usability, extensibility, and multi-GPU capability.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What interfaces and configuration options does AdvSecureNet provide?",{"text":79,"@type":75},"AdvSecureNet supports both command-line interface (CLI) and application programming interface (API). It also enables external YAML configuration files to improve versatility and reproducibility.",{"name":81,"@type":72,"acceptedAnswer":82},"How does AdvSecureNet evaluate adversarial robustness and attack effectiveness?",{"text":83,"@type":75},"It provides evaluation metrics including accuracy, robustness, transferability, and similarity. Robustness measures resistance to attacks, while transferability assesses deception across different models and similarity uses perceptual measures such as PSNR and SSIM.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":91},[92,96,100,104,109,112,117,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":110,"slug":111},50,"technology",{"id":113,"doc_module":4,"doc_module_name":45,"category_name":114,"show_sort_weight":115,"slug":116},7,"Healthcare",40,"healthcare",{"id":118,"doc_module":4,"doc_module_name":45,"category_name":119,"show_sort_weight":120,"slug":121},8,"Research & Report",30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]