[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118537-en":3,"doc-seo-118537-105":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},118537,962075114101,"Seraphina","https://ap-avatar.wpscdn.com/avatar/e000253a75eb197efd?x-image-process=image/resize,m_fixed,w_180,h_180&k=1780044092746381165",8,"Research & Report","A Survey on Zero Knowledge Proofs and their applications on Machine Learning","Machine Learning (ML) is increasingly used in both critical and non-critical services, but providers often keep model details private. This prevents confidence in whether results were produced using the declared model, verification of claimed model properties, and assurance that the production process followed agreed rules. Zero Knowledge Proofs (ZKP) offer a way to prove correct computation execution while preserving secrecy. Because ZKP supports general-purpose computations, it can be applied to ML as ZKML. The thesis analyzes core protocols, evaluates suitability for ML, provides quantitative performance analysis on ML models, reviews literature solutions, and discusses future directions.","A Survey on Zero Knowledge Proofs and their applications on Machine Learning  \nTesi di Laurea Magistrale in  \nComputer Science and Engineering - Ingegneria Informatica  \nAuthor: Luca Cerioli  \nStudent ID: 964191  \nAdvisor: Prof. Giovanni Ennio Quattrocchi  \nAcademic Year: 2021-22  \ni  \nAbstract  \nMachine Learning (ML) is a discipline that has seen increasing use in many services offered to users, both in critical processes, such as autonomous driving and personal health, and in less critical processes such as chat-bots or virtual assistants and recommendation systems. In many cases, the models used are owned by the service providers and therefore cannot be revealed to the public. This has important consequences, in fact: it is not possible to be certain that a result has been produced using the declared model from the data provided. Secondly, it is not possible to verify that the declared properties of a model are valid and present. Besides it is not possible to be certain that the production process of a model has followed certain rules. This is due to the secrecy of the model, which therefore prevents direct inspection. For this reason, Zero Knowledge Proof (ZKP) is a candidate as a plausible solution. This methodology makes it possible to obtain evidence on the actual execution of a computation, guaranteeing the secrecy of those elements that may not be revealed. Given that ZKP can be applied to all kinds of general-purpose computations, it follows that it is also possible to apply it to ML computations. Given the novelty of this combined approach (ZKP and ML), it is interesting to understand its feasibility and any critical points to focus on. Initially, this thesis will offer an analysis of ZKP by highlighting the main protocols and outlining characteristics that allow an assessment of their suitability for ML. Next, a quantitative analysis regarding the application of ZKP techniques on some ML models will be proposed to outline the performance of this approach and identify which approaches are the most promising. Afterwards, we present an analysis of the initial solutions proposed in the literature that attempt to apply ZKP to ML models. Last but not least, we will briefly discuss some future works and directions which can help in overcoming some of the problems we detected during our analysis.  \nKeywords: machine learning, zero-knowledge, ZKML, survey, verifiable machine learning  \nAbstract in lingua italiana  \nIl Machine Learning (ML) è una disciplina che ha visto un utilizzo sempre più crescente in molti servizi proposti agli utenti, sia in processi critici, come nel caso della guida autonoma e della salute personale, sia in processi meno critici come chatbot o assistenti virtuali esistemi di raccomandazione. In molti casi, i modelli utilizzati sono di proprietà dei fornitori dei servizi e quindi non possono essere svelati al pubblico. Questo ha dei riscontri importanti infatti: non è possibile essere certi che un risultato sia stato effettivamenteprodotto usando il modello dichiarato a partire dai dati forniti. In secondo luogo non è possibile verificare che le proprietà dichiarate di un modello siano effettivamente valide epresenti. Non è possibile nemmeno essere sicuri che il processo di produzione di un modello abbia seguito determinate regole. Tutto ciò a causa della segretezza del modello, che quindi impedisce di usare tecniche di ispezione in chiaro. Per questo motivo, Zero Knowledge Proof (ZKP) si candida come una plausibile soluzione. Infatti questa metodologia permette di ottenere delle prove sull’effettiva esecuzione di una computazione, garantendola segretezza degli elementi che non sono da tenere in chiaro. Dato che le computazioni su cui è possibile applicare ZKP sono di tipo general-purpose, ne consegue che è possibile percorrere questa strada anche per computazioni di ML. Data la novità di questo approccio combinato (ZKP e ML), è interessante capire quale sia la sua fattibilità ed eventuali punti criti","cbCaiu9NoezUdvwF","https://ap.wps.com/l/cbCaiu9NoezUdvwF","pdf",2195424,1,134,"English","en",105,"# 1 Introduction\n## 1.1 Research Problems\n## 1.2 Contributions\n## 1.3 Thesis Structure\n# 2 Background\n## 2.1 Verifiable Computing\n## 2.2 Zero Knowledge Proof\n## 2.3 Related Works\n# 3 Problem Statement\n## 3.1 Why ZKML\n## 3.2 Real use-case scenarios\n# 4 Theoretical Approaches\n## 4.1 Systematic Literature Review","[{\"question\":\"Why is verifiable machine learning needed in practice?\",\"answer\":\"Because ML models are often owned by service providers and kept private, it is difficult to confirm that outputs were produced with the declared model, or that stated model properties and production rules hold.\"},{\"question\":\"What role do Zero Knowledge Proofs play in ZKML?\",\"answer\":\"ZKP enables evidence about the actual execution of computations while keeping sensitive elements secret, making it possible to apply this idea to ML computations.\"},{\"question\":\"How is the thesis structured to evaluate ZKML?\",\"answer\":\"It first analyzes ZKP protocols for suitability to ML, then performs quantitative analysis of ZKP techniques on selected ML models, reviews current literature solutions, and concludes with future work directions.\"}]","A Survey on Zero Knowledge Proofs and their applications on Machine Learning | PDF",1785684038,338,{"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},"a-survey-on-zero-knowledge-proofs-and-their-applications-on-machine-learning","",{"@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/a-survey-on-zero-knowledge-proofs-and-their-applications-on-machine-learning/118537/",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","2026-08-02",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},"Why is verifiable machine learning needed in practice?","Question",{"text":75,"@type":76},"Because ML models are often owned by service providers and kept private, it is difficult to confirm that outputs were produced with the declared model, or that stated model properties and production rules hold.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What role do Zero Knowledge Proofs play in ZKML?",{"text":80,"@type":76},"ZKP enables evidence about the actual execution of computations while keeping sensitive elements secret, making it possible to apply this idea to ML computations.",{"name":82,"@type":73,"acceptedAnswer":83},"How is the thesis structured to evaluate ZKML?",{"text":84,"@type":76},"It first analyzes ZKP protocols for suitability to ML, then performs quantitative analysis of ZKP techniques on selected ML models, reviews current literature solutions, and concludes with future work directions.","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"]