[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120860-en":3,"doc-seo-120860-105":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},120860,1649267921044,"Ava Thompson","https://us-avatar.wpscdn.com/avatar/1800007509477c92dfb?_k=1782875107921204101",8,"Research & Report","Multi-party Computation for Privacy and Security in Machine Learning - a Practical Review","Machine learning, especially deep learning, drives advances across critical domains such as healthcare, finance, transportation, and education, yet it introduces significant privacy and security weaknesses throughout the lifecycle. These risks stem from the approximate and attackable nature of ML models, the pervasive exposure of datasets and model parameters, and the sensitivity of training procedures to manipulation. Multi-party computation (MPC) is presented as a practical family of techniques to mitigate ML privacy and security threats by enabling collaborative model evaluation without revealing private data.","POLITECNICO DI TORINO  \nRepository ISTITUZIONALE  \nMulti-party Computation for Privacy and Security in Machine Learning: a Practical Review  \nOriginal  \nMulti-party Computation for Privacy and Security in Machine Learning: a Practical Review / Bellini, Alessandro; Bellini, Emanuele; Bertini, Massimo; Almhaithawi, Doaa; Cuomo, Stefano. - (2023), pp. 174-179. (Intervento presentato al convegno 2023 IEEE International Conference on Cyber Security and Resilience (CSR) tenutosi a Venice (Italy) nel 31 July 2023-02 August 2023) [10 . 1109/CSR57506 .2023. 10224826] .  \nAvailability:  \nThis version is available at: 11583/2982820 since: 2023-10-19T10:40:52Z  \nPublisher: IEEE  \nPublished  \nDOI:10.1109/CSR57506.2023.10224826  \nTerms of use:  \nThis article is made available under terms and conditions as specified in the corresponding bibliographic description in the repository  \nPublisher copyright  \n(Article begins on next page)  \n18 September 2024  \nMulti-party Computation for Privacy and Security in Machine Learning: a practical review  \nAlessandro Bellini  \nMathema s.r.l. Florence, Italy [Email:abel@mathema.com](Email:abel@mathema.com)  \nEmanuele Bellini  \nLOGOS-RI Florence, Italy  \nEmail: [emanuele.bellini@logos-ri.eu](emanuele.bellini@logos-ri.eu)[ ](emanuele.bellini@logos-ri.eu)[ORCID:0000-0002-7878-8710](ORCID:0000-0002-7878-8710)  \nMassimo Bertini  \nMathema s.r.l.  \nFlorence, Italy  \nEmail: [stefano.cuomo@mathema.com](stefano.cuomo@mathema.com)  \nDoaa Almhaithawi  \nDepartment of Control and Computer Engineering Politecnico di Torino Torino, Italy Email: Doaa.Almhaithawi@polito.it  \nStefano Cuomo  \nMathema s.r.l.  \nFlorence, Italy  \nEmail: [stafano.cuomo@mathema.com](stafano.cuomo@mathema.com)  \nAbstract—Machine Learning, particularly Deep Learning, is transforming society in any of its fundamental domains-healthcare, culture, finance, transportation, education, just to mention a few. However Machine Learning suffers from serious weaknesses in privacy and security due to the large amount of data (datasets for training and parameters in trained models) and the probabilistic approximation inherent in any ML function. MultiParty Computation (MPC) is a family of techniques and tactic with a sound scientific and operative base that can be applied to mitigate some relevant weaknesses of ML. In particular, privacy in training may be assured by MPC with federated learning techniques (these may be considered particular interpretationsand implementation of a general MPC method) and also security in training and inference may be enforced by continuous model testing using MPC is a technique that allows multiple parties to evaluate a machine learning model on their private data without revealing it to each other. This brief paper is a practical and essential review on how to use MPC to mitigate privacy and security issues in ML.  \nI. INTRODUCTION  \nMachine Learning (ML), particularly Deep Learning, hasan impressive and an unprecedented success in a wide range of applications encompassing any crucial domain of human activity such as healthcare [1], , finance [2], retail [3], transportation [4], education [5], entertainment [6], manufacturing [7] . However ML poses a serious security menace along the whole life-cycle from dataset, through training, to inference. Substantially, ML consists of computing the parameters (weights) of an approximate function by an optimization process on a dataset. So evident potential surface of attack is represented by:  \n• The approximate nature of ML functions can be intrinsically fooled (it is an approximation);  \n• Data are pervasive (dataset and parameters) posing both privacy and security issues;  \n• The optimization process if only slightly altered may compromise the generation of the ML approximate function.  \nA. Types of Attack on Machine Learning  \nML is subjected to several attacks in the training and in the inference phases. Among the most relevant are:  \n• Poisoning: Poisoning attacks involve injecting ","cbCaiazTzdgxo3Fv","https://ap.wps.com/l/cbCaiazTzdgxo3Fv","pdf",192498,1,7,"English","en",105,"# Introduction\n## Types of Attack on Machine Learning\n# Multi-party Computation as a Mitigation Approach","[{\"question\":\"Why does machine learning face privacy and security threats across its lifecycle?\",\"answer\":\"Because ML models are approximate and can be fooled, datasets and model parameters are pervasive, and training can be compromised by even slight changes to the optimization process.\"},{\"question\":\"What are the main categories of attacks discussed for machine learning?\",\"answer\":\"The document highlights poisoning (malicious training data), adversarial examples (perturbed inputs causing wrong predictions), and model inversion (inferring sensitive training information from model outputs).\"},{\"question\":\"How does multi-party computation (MPC) help mitigate these privacy and security issues?\",\"answer\":\"MPC enables multiple parties to evaluate an ML model on private data without revealing that data to each other, supporting privacy in training (including with federated learning techniques) and security through continuous model testing using MPC.\"}]","Multi-party Computation for Privacy and Security in Machine Learning - a Practical Review | PDF",1785732386,18,{"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},"multi-party-computation-for-privacy-and-security-in-machine-learning-a-practical-review","",{"@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/multi-party-computation-for-privacy-and-security-in-machine-learning-a-practical-review/120860/",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},"Why does machine learning face privacy and security threats across its lifecycle?","Question",{"text":75,"@type":76},"Because ML models are approximate and can be fooled, datasets and model parameters are pervasive, and training can be compromised by even slight changes to the optimization process.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What are the main categories of attacks discussed for machine learning?",{"text":80,"@type":76},"The document highlights poisoning (malicious training data), adversarial examples (perturbed inputs causing wrong predictions), and model inversion (inferring sensitive training information from model outputs).",{"name":82,"@type":73,"acceptedAnswer":83},"How does multi-party computation (MPC) help mitigate these privacy and security issues?",{"text":84,"@type":76},"MPC enables multiple parties to evaluate an ML model on private data without revealing that data to each other, supporting privacy in training (including with federated learning techniques) and security through continuous model testing using MPC.","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},[93,97,101,105,110,115,119,122,127,130,134],{"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":21,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]