[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-124896-en":3,"doc-seo-124896-105":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":20,"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},124896,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","In-Network Machine Learning for Real-Time Transaction Fraud Detection","Machine learning enables more intelligent transaction-fraud prevention, but real-time detection must achieve ultra-low latency to flag suspicious activity before completion. The work introduces MIND, an ML-based fraud-detection approach that performs inference inside programmable network devices to reduce delay while still mitigating fraudulent transactions. MIND is prototyped on software and hardware platforms and evaluated on three public datasets. Results show microsecond-scale latency, high throughput, and accuracy and F1-score close to server-based benchmarks, reducing servers, costs, and energy while improving customer experience.","2902  \nECAI 2024  \nU. Endriss et al. (Eds.)© 2024 The Authors.  \nThis article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0).  \ndoi:10.3233/FAIA240828  \nIn-Network Machine Learning for Real-Time Transaction Fraud Detection  \nXinpeng Honga, * , Changgang Zhenga and Noa ZilbermanaaDepartment of Engineering Science, University of Oxford, Oxford, United Kingdom  \nAbstract. Machine learning (ML) has become a mainstream approach in the fight against transaction fraud for its intelligence. For financial institutions and businesses, low-latency detection of fraudulent transactions in real-time is highly important as it enables rapid identification and prevention. Concurrently mitigating fraudulent transactions by using ML while also reducing latency remains a challenging endeavor, for which performing inference within programmable network devices offers a potential solution. In this paper, we introduce MIND, conducting ML-based fraud detection within programmable devices. MIND is prototyped on both software and hardware network devices, including BMv2, Intel Tofino, and NVIDIA BlueField-2 DPU, and is evaluated with three publicly available transaction datasets. Experimental results demonstrate that MIND detects transaction fraud in real-time, with a throughput of 6.4 terabits per second and microsecond-scale latency. Compared with server-based solutions, MIND can process over ×800 more transactions per second, along with a latency reduction of over × 1300 per transaction. At the same time, MIND attains 99.94% of serverbased benchmarks’ accuracy and 93.66% of their F1-score, exhibiting only marginal degradation in classification performance. Therefore, MIND offers substantial savings in the number of servers, leading to reduced costs and energy consumption, while providing a better customer experience.  \n1 Introduction  \nFraudulent activities are widespread in financial transactions nowadays. These refer to any intentional deception or misrepresentation by an individual or group with the aim of obtaining unauthorized benefits [38] . Transaction fraud can have significant and wideranging harmful effects on individuals, businesses, and society as a whole. According to a study by Juniper Research, merchants’ cumulative financial losses from online transaction fraud are predicted to surpass $343 billion between 2023 and 2027 [30] . Therefore, fraud detection plays a crucial role in preventing and reducing the possibility of potential harm. Fraud Detection Systems (FDS) are designed to identify and flag suspicious fraudulent behaviors, allowing for prompt intervention and prevention of further damage [2] . However, the ever-evolving nature of fraudulent activities, coupled with fraudsters’ ability to adapt to fraud detection measures, presents a considerable challenge to traditional FDS with statistical and rulebased analysis components [46] .  \nTo better secure electronic commerce systems, most FDS utilize machine learning (ML) algorithms to recognize fraudulent patterns  \n∗ Corresponding Author. Email: [xinpeng.hong@eng.ox.ac.uk](xinpeng.hong@eng.ox.ac.uk).  \nFigure 1. General working scenario of MIND.  \nand detect them in real-time transaction data streams [6] . ML has significantly improved detection accuracy and reduced false positives thanks to its ability to analyze complex datasets and learn from anomalies in historical transactions [4] . Despite the promising results demonstrated by previous ML-based works, accurate and prompt detection remains a formidable task. Significant data imbalance and considerable variability of fraudulent transactions both contribute to the complexity [14, 33] . To reach optimal performance, there is an ongoing development of increasingly advanced models. However, embedding more sophisticated ML models into FDS conflicts with the aim of lowering detection time, which is essential to prevent fraudulent trans","cbCais1MFbN4lQgQ","https://ap.wps.com/l/cbCais1MFbN4lQgQ","pdf",764895,1,"English","en",105,"# Introduction\n## Real-time fraud detection challenges\n## Why in-network ML and programmable devices\n## MIND approach and deployment scenario","[{\"question\":\"What problem does MIND address in transaction fraud detection?\",\"answer\":\"MIND targets the challenge of detecting fraudulent transactions in real time with extremely low latency, where traditional systems using server-side processing can be too slow for microsecond-level threats.\"},{\"question\":\"How does MIND perform fraud detection differently from server-based systems?\",\"answer\":\"MIND runs ML inference within programmable network devices (in the data plane), so suspicious transactions are flagged inside the switch and avoid server processing and related delay.\"},{\"question\":\"What evaluation results does MIND report?\",\"answer\":\"MIND prototypes achieve real-time detection with microsecond-scale latency and very high throughput, while maintaining accuracy and F1-score close to server-based benchmarks with only marginal degradation.\"}]","In-Network Machine Learning for Real-Time Transaction Fraud Detection | PDF",1785895278,20,{"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},"in-network-machine-learning-for-real-time-transaction-fraud-detection","",{"@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/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/in-network-machine-learning-for-real-time-transaction-fraud-detection/124896/",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","2026-08-05",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does MIND address in transaction fraud detection?","Question",{"text":74,"@type":75},"MIND targets the challenge of detecting fraudulent transactions in real time with extremely low latency, where traditional systems using server-side processing can be too slow for microsecond-level threats.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does MIND perform fraud detection differently from server-based systems?",{"text":79,"@type":75},"MIND runs ML inference within programmable network devices (in the data plane), so suspicious transactions are flagged inside the switch and avoid server processing and related delay.",{"name":81,"@type":72,"acceptedAnswer":82},"What evaluation results does MIND report?",{"text":83,"@type":75},"MIND prototypes achieve real-time detection with microsecond-scale latency and very high throughput, while maintaining accuracy and F1-score close to server-based benchmarks with only marginal degradation.","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},[92,96,100,104,109,114,119,122,126,129,133],{"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":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]