[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121028-en":3,"doc-seo-121028-105":30,"detail-sidebar-cat-0-en-105":91},{"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":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121028,5909877438554,"Maeve","https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272",8,"Research & Report","Hawk - Accurate and Fast Privacy-Preserving Machine Learning Using Secure Lookup Table Computation","Training machine learning models across multiple data owners without direct data sharing enables applications blocked by business, legal, or ethical constraints. This work designs privacy-preserving protocols for logistic regression and neural networks using a two-server secret-sharing model. It targets the inefficiency and inaccuracy of Yao’s garbled circuits for non-linear activations by using secure lookup tables. It also studies relaxed security, quantifying leakage and proving a form of X-privacy for access patterns, improving efficiency substantially.","Hawk: Accurate and Fast Privacy-Preserving Machine Learning Using Secure Lookup Table Computation  \nHamza Saleem  \nUniversity of Southern California [hsaleem@usc](hsaleem@usc.edu)[.](hsaleem@usc.edu)[edu](hsaleem@usc.edu)  \nMuhammad Naveed  \nUniversity of Southern California [mnaveed@usc](mnaveed@usc.edu)[.](mnaveed@usc.edu)[edu](mnaveed@usc.edu)  \nAmir Ziashahabi  \nUniversity of Southern California [ziashaha@usc](ziashaha@usc.edu)[.](ziashaha@usc.edu)[edu](ziashaha@usc.edu)  \nSalman Avestimehr  \nUniversity of Southern California [avestime@usc](avestime@usc.edu)[.](avestime@usc.edu)[edu](avestime@usc.edu)  \narXiv :2403 . 17296v1 [ cs .CR] 26 Mar 2024  \nABSTRACT  \nTraining machine learning models on data from multiple entities without direct data sharing can unlock applications otherwise hindered by business, legal, or ethical constraints. In this work, we design and implement new privacy-preserving machine learning protocols for logistic regression and neural network models. We adopt a two-server model where data owners secret-share their data between two servers that train and evaluate the model on the joint data. A significant source of inefficiency and inaccuracy in existing methods arises from using Yao’s garbled circuits to compute non-linear activation functions. We propose new methods for computing non-linear functions based on secret-shared lookup tables, offering both computational efficiency and improved accuracy.  \nBeyond introducing leakage-free techniques, we initiate the exploration of relaxed security measures for privacy-preserving machine learning. Instead of claiming that the servers gain no knowledge during the computation, we contend that while some information is revealed about access patterns to lookup tables, it maintains 􀁮-􀀳X-privacy. Leveraging this relaxation significantly reduces the computational resources needed for training. We present new cryptographic protocols tailored to this relaxed security paradigm and define and analyze the leakage. Our evaluations show that our logistic regression protocol is up to 9× faster, and the neural network training is up to 688× faster than SecureML [58] . Notably, our neural network achieves an accuracy of 96. 6% on MNIST in 15 epochs, outperforming prior benchmarks [58, 76] that capped at 93.4% using the same architecture.  \nKEYWORDS  \nsecure multi-party computation, privacy-preserving ML  \n1 INTRODUCTION  \nMachine learning (ML) has become an indispensable tool in various domains, from advertising and healthcare to finance and retail, enabling the training of predictive models. The accuracy of these models is significantly enhanced when trained on vast datasets aggregated from diverse sources. However, numerous challenges, such as privacy concerns, stringent regulations, and competitive  \nThis work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license visit [https://creativecommons](https://creativecommons.org/licenses/by/4.0/ or)[.](https://creativecommons.org/licenses/by/4.0/ or)[org/licenses/by/4](https://creativecommons.org/licenses/by/4.0/ or)[.](https://creativecommons.org/licenses/by/4.0/ or)[0/ or](https://creativecommons.org/licenses/by/4.0/ or) send a  \nletter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. Proceedings on Privacy Enhancing Technologies YYYY(X), 1–17  \n© YYYY Copyright held by the owner/author(s) . [https://doi](https://doi.org/XXXXXXX.XXXXXXX)[.](https://doi.org/XXXXXXX.XXXXXXX)[org/XXXXXXX](https://doi.org/XXXXXXX.XXXXXXX)[.](https://doi.org/XXXXXXX.XXXXXXX)[XXXXXXX](https://doi.org/XXXXXXX.XXXXXXX)  \nbusiness landscapes, often hinder collaborative data-sharing endeavors.  \nPrivacy-Preserving Machine Learning (PPML) using secure computation allows multiple entities to train models on their aggregated data without revealing their individual datasets. In this paradigm, data owners distribute their private data among several noncolluding servers using secret-sharing ","cbCainW8KdKgeGBD","https://ap.wps.com/l/cbCainW8KdKgeGBD","pdf",1248270,1,17,"English","en",105,"# Abstract\n# Introduction\n## Privacy-preserving machine learning background\n## Bottlenecks in existing PPML approaches\n## Hawk Single and Hawk Multi protocols\n## Two-server PPML protocols for logistic regression and neural networks","[{\"question\":\"What problem does Hawk address in privacy-preserving machine learning?\",\"answer\":\"It addresses the inefficiency and accuracy loss caused by using Yao’s garbled circuits to compute non-linear activation functions in existing PPML methods.\"},{\"question\":\"How do the Hawk Single and Hawk Multi protocols differ?\",\"answer\":\"Hawk Single provides plaintext-accuracy computation using secret-shared lookup tables but consumes one table per function computation. Hawk Multi enables lookup table reusability while introducing and analyzing additional leakage from access patterns.\"},{\"question\":\"What performance and accuracy results does the paper report?\",\"answer\":\"The logistic regression protocol is up to 9× faster, and neural network training is up to 688× faster than SecureML. The neural network reaches 96.6% accuracy on MNIST in 15 epochs, exceeding prior benchmarks using the same architecture.\"}]","Hawk - Accurate and Fast Privacy-Preserving Machine Learning Using Secure Lookup Table Computation | PDF",1785733388,43,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"hawk-accurate-and-fast-privacy-preserving-machine-learning-using-secure-lookup-table-computation","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/hawk-accurate-and-fast-privacy-preserving-machine-learning-using-secure-lookup-table-computation/121028/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does Hawk address in privacy-preserving machine learning?","Question",{"text":75,"@type":76},"It addresses the inefficiency and accuracy loss caused by using Yao’s garbled circuits to compute non-linear activation functions in existing PPML methods.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How do the Hawk Single and Hawk Multi protocols differ?",{"text":80,"@type":76},"Hawk Single provides plaintext-accuracy computation using secret-shared lookup tables but consumes one table per function computation. Hawk Multi enables lookup table reusability while introducing and analyzing additional leakage from access patterns.",{"name":82,"@type":73,"acceptedAnswer":83},"What performance and accuracy results does the paper report?",{"text":84,"@type":76},"The logistic regression protocol is up to 9× faster, and neural network training is up to 688× faster than SecureML. The neural network reaches 96.6% accuracy on MNIST in 15 epochs, exceeding prior benchmarks using the same architecture.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]