[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117109-en":3,"doc-seo-117109-105":30,"detail-sidebar-cat-0-en-105":92},{"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":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},117109,4398048949847,"Eliana","https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267",8,"Research & Report","In-Network Machine Learning Using Programmable Network Devices - A Survey","Machine learning tackles core networking problems such as traffic classification, anomaly detection, and network configuration, yet its computation can strain devices, networks, and servers. In-network computing, enabled by programmable network devices, runs applications inside the network to improve throughput and reduce latency. This survey presents in-network machine learning, offering a comprehensive taxonomy, describing technology fundamentals, and detailing solution types and model implementations within programmable network devices. It also examines challenges, benefits for cloud and next-generation networks, and future trends.","In-Network Machine Learning Using Programmable  \nNetwork Devices: A Survey  \nChanggang Zheng , Xinpeng Hong , Damu Ding , Shay Vargaftik , Yaniv Ben-Itzhak ,  \nNoa Zilberman , Senior Member, IEEE  \nAbstract—Machine learning is widely used to solve networking challenges, ranging from traffic classification and anomaly detection to network configuration. However, machine learning also requires significant processing and often increases the load on both networks and servers. The introduction of in-network computing, enabled by programmable network devices, has allowed to run applications within the network, providing higher throughput and lower latency. Soon after, in-network machine learning solutions started to emerge, enabling machine learning functionality within the network itself.  \nThis survey introduces the concept of in-network machine learning and provides a comprehensive taxonomy. The survey provides an introduction to the technology and explains the different types of machine learning solutions built upon programmable network devices. It explores the different types of machine learning models implemented within the network, and discusses related challenges and solutions. In-network machine learning can significantly benefit cloud computing and nextgeneration networks, and this survey concludes with a discussion of future trends.  \nIndex Terms—In-network computing; Machine learning; P4; Programmable data planes; Software Defined Networks.  \nI. INTRODUCTION  \nC  \nLOUD and edge computing are attending to the increasing flow  \nbecoming powerful, of data from users  \nto cloud-based services. The new generation of networkprocessing technology revolutionizes network infrastructure as we know it and supports the increasing demand for networktraffic forwarding and processing.  \nRecent network devices are no longer just highperformance, but also programmable. Switch-ASIC (e.g., [1, 2]), network interface cards (NICs) [3, 4, 5, 6], and FPGAbased network devices [7] use a domain-specific language, P4 [8], to define and customize network protocols directly in the data plane. This programmability enables executing advanced network functions, and improves resources’ utilization [9, 10] . This brings new opportunities to offload computations and applications to network devices: computing entirely within a programmable network devices is called Innetwork Computing.  \nChanggang Zheng, Xinpeng Hong, Damu Ding, and Noa Zilberman are with the Computing Infrastructure Group, Department of Engineering Science, University of Oxford (e-mail: changgang.zheng, xinpeng.hong, damu.ding, [noa.zilberman@eng.ox.ac.uk](noa.zilberman@eng.ox.ac.uk))  \nShay Vargaftik, and Yaniv Ben-Itzhak are with the VMware (by Broadcom) Research Group (e-mail: [shayv@vmware.com](shayv@vmware.com), yaniv.benitzhak@broadcom.com)  \nMachine learning (ML) was shown long ago to be useful for traffic classification [11, 12] and for network anomaly detection [13] . These network-oriented ML tasks are typically deployed on servers or middleboxes [14] . However, as data volume increases so do the processing demands from the devices running the ML-based tasks.  \nThe popularity of ML for networking, and the rising packetprocessing demands have led to the suggestion that running ML algorithms on programmable network devices can significantly improve ML performance in terms of throughput and latency [15, 16] . Furthermore, it can help reduce memory consumption and communication overheads [17] . Consequently, a wide range of ML algorithms have been implemented indifferent ways on multiple types of programmable network devices.  \nIn this survey, we distinguish between three forms of ML execution: General ML, Network-Assisted ML, and InNetwork ML. General ML refers to the case where both the ML model training and decision making are on the server side, including deployments on hardware accelerators such as GPU. Network-Assisted ML uses network devices primarily for model training acce","cbCaivMrcsbBS0fV","https://ap.wps.com/l/cbCaivMrcsbBS0fV","pdf",5783687,1,35,"English","en",105,"# Introduction\n## In-network computing and programmability\n## ML for networking and deployment placement\n## Three forms of ML execution\n# Survey scope and contributions","[{\"question\":\"What problem does in-network machine learning address in conventional networking ML?\",\"answer\":\"Conventional ML for networking increases processing demands on servers and network devices, raising load, latency, and overheads. In-network ML offloads computation into programmable devices to improve throughput and reduce latency.\"},{\"question\":\"How do General ML, Network-Assisted ML, and In-Network ML differ?\",\"answer\":\"General ML keeps both training and decision making on the server side. Network-Assisted ML uses network devices for training acceleration and better feature collection while inference remains on the end host. In-Network ML performs complete ML processes, either training or inference, entirely within the network.\"},{\"question\":\"What role do programmable network devices and P4 play in enabling in-network ML?\",\"answer\":\"Programmable devices use a domain-specific language, P4, to define and customize network protocols in the data plane. This programmability enables advanced network functions and provides opportunities to execute ML-based tasks within the network.\"}]","In-Network Machine Learning Using Programmable Network Devices - A Survey | PDF",1785673841,88,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":28},"in-network-machine-learning-using-programmable-network-devices-a-survey","",{"@graph":36,"@context":86},[37,54,69],{"@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/in-network-machine-learning-using-programmable-network-devices-a-survey/117109/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05","2026-08-02",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What problem does in-network machine learning address in conventional networking ML?","Question",{"text":76,"@type":77},"Conventional ML for networking increases processing demands on servers and network devices, raising load, latency, and overheads. In-network ML offloads computation into programmable devices to improve throughput and reduce latency.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How do General ML, Network-Assisted ML, and In-Network ML differ?",{"text":81,"@type":77},"General ML keeps both training and decision making on the server side. Network-Assisted ML uses network devices for training acceleration and better feature collection while inference remains on the end host. In-Network ML performs complete ML processes, either training or inference, entirely within the network.",{"name":83,"@type":74,"acceptedAnswer":84},"What role do programmable network devices and P4 play in enabling in-network ML?",{"text":85,"@type":77},"Programmable devices use a domain-specific language, P4, to define and customize network protocols in the data plane. This programmability enables advanced network functions and provides opportunities to execute ML-based tasks within the network.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]