[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290614-105":59,"doc-detail-290614-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","recent-advances-of-privacy-preserving-machine-learning-based-on-fully-homomorphic-encryption-recent-research-overview","Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption - Recent research overview","","Fully Homomorphic Encryption (FHE) enables computation on encrypted data without decryption, offering a strong foundation for privacy-preserving machine learning (PPML). The document notes that numerous FHE schemes and many FHE-based PPML solutions exist, and they are evolving rapidly, making it hard to form a complete view. It introduces representative FHE-based PPML results, compares the practical trade-offs of different approaches, and aims to help users select methods matched to their needs and expected efficiency.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/recent-advances-of-privacy-preserving-machine-learning-based-on-fully-homomorphic-encryption-recent-research-overview/290614/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/recent-advances-of-privacy-preserving-machine-learning-based-on-fully-homomorphic-encryption-recent-research-overview/290614.png","ImageObject",300,407,{"name":92,"@type":93},"Ezra","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-28","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",14,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What does Fully Homomorphic Encryption enable for machine learning over encrypted data?","Question",{"text":113,"@type":114},"It allows data manipulation and computation in the encrypted domain without decryption, making privacy-preserving machine learning training or inference possible.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does the document categorize different FHE schemes?",{"text":118,"@type":114},"It follows a four-generation classification: first-generation (theoretical but too slow), second-generation (often SIMD-capable and efficient), third-generation (faster bootstrapping but hard to support SIMD), and fourth-generation (approximate schemes such as CKKS for efficient approximate arithmetic).",{"name":120,"@type":111,"acceptedAnswer":121},"What is the purpose of the article’s overview of FHE-based PPML research?",{"text":122,"@type":114},"It helps readers identify candidate methods for their use case, understand pros and cons among approaches, and decide whether available works are mature enough for their efficiency and requirements.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},290614,1789642198,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":34,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":130,"read_time":144,"preview_page_count":4},1099514068035,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","Security and Safety, Vol. 4, 2024012 (2025)  \n[https://doi.org/10.1051/sands/2024012](https://doi.org/10.1051/sands/2024012)[ ](https://doi.org/10.1051/sands/2024012)[sands.edpsciences.org](sands.edpsciences.org)  \nSecurity and Safety of Data in Cloud Computing  \nOther Fields  \nRecent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption  \nCheng Hong􀀃 Ant Group, Beijing 100081, China  \nReceived: 31 July 2024 / Revised: 9 September 2024 / Accepted: 10 September 2024 / Published online: 18 October 2024  \nAbstract Fully Homomorphic Encryption (FHE), known for its ability to process encrypted data without decryption, is a promising technique for solving privacy concerns in the machine learning era. However, there are many kinds of available FHE schemes and way more FHEbased solutions in the literature, and they are still fast evolving, making it di􀀎cult to geta complete view. This article aims to introduce recent representative results of FHE-based privacy-preserving machine learning, helping users understand the pros and cons of di􀀋erent  \nkinds of solutions, and choose an appropriate approach for their needs.  \nKeywords Homomorphic Encryption, Fully Homomorphic Encryption, Machine learning,  \nPrivacy-preserving machine learning  \nCitation Hong C. Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption. Security and Safety 2025; 4: 2024012. [https://doi.org/10.1051/](https://doi.org/10.1051/)  \n[sands/2024012](sands/2024012)  \n1 Introduction  \nFully Homomorphic Encryption (FHE) is a technology that allows data manipulation in the encrypted domain without decryption. The idea of FHE was 􀀌rst introduced by Rivest et al. [1] in 1978, but it was not until 2008 that the 􀀌rst FHE scheme [2] was proposed by Gentry. FHE was considered just theoretic and impractical in its early ages, but today many di􀀋erent FHE schemes [3{7] have been proposed, with their e􀀎ciency greatly improved.  \nOne of the promising applications of FHE is that it could enable privacy-preserving machine learning (PPML) training or inference on encrypted data, protecting data con􀀌dentiality and privacy. There already exist many research works that use FHE to build PPML solutions, some of them even come close to the level of industrial deployments. Given di􀀋erent types of FHE schemes and machine learning tasks, the number of possible combinations is vast, and the area is still fast evolving, thus it's often di􀀎cult to answer the question below without enough investigation:  \nQ: Suppose I want to do privacy-preserving machine learning training/inference of model X on data Y using FHE. Are there any candidate methods for my use case? If yes, which one should I choose? What level of e􀀎ciency should I expect?  \nThis article tries to help users answer the question by providing a brief view of the recent research progresses of FHE-based PPML, help them choose research works to follow up, or decide whether the works are mature enough for their needs.  \n*  \nCorresponding author (email: [vince.hc@antgroup.com](vince.hc@antgroup.com))  \nThis is an Open Access article distributed under the terms of the Creative Commons Attribution License ([https://creativecommons.org/licenses/by/4.0](https://creativecommons.org/licenses/by/4.0)), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.  \n© The Author(s) 2025 . Published by EDP Sciences and China Science Publishing & Media Ltd.  \nSecurity and Safety, Vol. 4, 2024012  \nFigure 1 . Representative FHE schemes and applications. Numbers in small boxes refer to the generation of FHE, and `M'refers to mixed-mode FHE. Small boxes with white backgrounds refer to FHE schemes, and dark backgrounds refer to FHE applications  \nAccording to a talk by Gentry [8], existing FHE schemes could be divided into four generations as follows:  \n(1) The 􀀌rst generation FHE refers to Gentry's original design [2] based ","cbCair1tyslMRpvV","https://ap.wps.com/l/cbCair1tyslMRpvV","pdf",930473,"English","# Introduction\n## Generations of FHE\n# Second-generation FHE applications\n## CryptoNets","[{\"question\":\"What does Fully Homomorphic Encryption enable for machine learning over encrypted data?\",\"answer\":\"It allows data manipulation and computation in the encrypted domain without decryption, making privacy-preserving machine learning training or inference possible.\"},{\"question\":\"How does the document categorize different FHE schemes?\",\"answer\":\"It follows a four-generation classification: first-generation (theoretical but too slow), second-generation (often SIMD-capable and efficient), third-generation (faster bootstrapping but hard to support SIMD), and fourth-generation (approximate schemes such as CKKS for efficient approximate arithmetic).\"},{\"question\":\"What is the purpose of the article’s overview of FHE-based PPML research?\",\"answer\":\"It helps readers identify candidate methods for their use case, understand pros and cons among approaches, and decide whether available works are mature enough for their efficiency and requirements.\"}]","Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption - Recent research overview | PDF",18]