[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117104-en":3,"doc-seo-117104-105":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},117104,4398048949847,"Eliana","https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267",8,"Research & Report","Planter - Rapid Prototyping of In-Network Machine Learning Inference","In-network machine learning inference enables high throughput and low latency by placing inference logic inside the network. Its deployment can improve application performance and energy efficiency, yet research adoption faces a high barrier due to the need for expertise in programmable data planes, machine learning, and domain-specific knowledge. Existing approaches are often one-time, difficult to reproduce, adapt, or port across platforms. This paper introduces Planter, a modular open-source framework for rapid prototyping across platforms and pipeline architectures.","Planter: Rapid Prototyping of In-Network Machine Learning Inference  \nChanggang Zheng†, Mingyuan Zang§ , Xinpeng Hong†, Liam Perreault†, Riyad Bensoussane†,  \nShay Vargaftik⋄ , Yaniv Ben-Itzhak⋄, and Noa Zilberman†  \n†University of Oxford,§ Technical University of Denmark, ⋄ VMware Research (by Broadcom){changgang.zheng, xinpeng.hong, [noa.zilberman](noa.zilberman}@eng.ox.ac.uk)[}](noa.zilberman}@eng.ox.ac.uk)[@eng.ox.ac.uk](noa.zilberman}@eng.ox.ac.uk), [minza@dtu.dk](minza@dtu.dk), {liam.perreault,  \n[riyad.bensoussane](riyad.bensoussane}@worc.ox.ac.uk)[}](riyad.bensoussane}@worc.ox.ac.uk)[@worc.ox.ac.uk](riyad.bensoussane}@worc.ox.ac.uk), {shay.vargaftik, yaniv.ben-itzhak}@broadcom.com  \nABSTRACT  \nIn-network machine learning inference provides high throughput and low latency. It is ideally located within the network, power efficient, and improves applications’ performance. Despite its advantages, the bar to in-network machine learning research is high, requiring significant expertise in programmable data planes, in addition to knowledge of machine learning and the application area. Existing solutions are mostly one-time efforts, hard to reproduce, change, or port across platforms. In this paper, we present Planter: a modular and efficient open-source framework for rapid prototyping of in-network machine learning models across a range of platforms and pipeline architectures. By identifying general mapping methodologies for machine learning algorithms, Planter introduces new machine learning mappings and improves existing ones. It provides users with several example use cases and supports different datasets, and was already extended by users to new fields and applications. Our evaluation shows that Planter improves machine learning performance compared with previous model-tailored works, while significantly reducing resource consumption and co-existing with network functionality. Planter-supported algorithms run at line rate on unmodified commodity hardware, providing billions of inference decisions per second.  \nKEYWORDS  \nIn-Network Computing; Machine Learning; Dimension Reduction; Modular Framework; Machine Learning Compilers; Programmable Switches; P4 .  \n1 INTRODUCTION  \nThe rapid growth of data volume and the increasing demands for data exploitation are creating an ever-increasing processing burden on computing systems [96] . The need to scale computing resources and the emergence of programmable network devices drove researchers to leverage the underused processing resources within the network [18, 41] . Processing within the network, also known as in-network computing, was shown to boost performance, while at the same time increasing power efficiency [84] . It was demonstrated to improve a range of applications, from network services and monitoring [3, 42, 45, 55] to caching and consensus [18, 41] .  \nIn-network computing was suggested as a means to improve machine learning (ML) performance, both in the acceleration of host-based ML training using in-network aggregation [29, 48, 73], and ML inference by using in-network classification [72, 91, 98] . In-network ML inference benefits from its deployment location,  \nProgrammable Network Devices  \nFigure 1: Difference in traffic flow between traditional ML in the network domain and in-network ML.  \nillustrated in Figure 1, as data can travel directly from source to destination, without additional hops through inference servers. This provides latency benefits and high throughput, without introducing additional traffic into the network, as in server or GPU-based inference.  \nWhile in-network ML is appealing, its practical adoption within the research community was limited for two reasons. First, from the design and realization perspective, there was no general solution or a standard methodology for mapping different ML models to programmable data planes. Furthermore, many mapping solutions suffered from stage and memory explosion [7, 26], thereby limiting model size and co","cbCaiiCAcCb7iYzE","https://ap.wps.com/l/cbCaiiCAcCb7iYzE","pdf",2706961,1,18,"English","en",105,"# Introduction\n## Motivation for in-network ML inference\n## Challenges in current research efforts\n## Planter framework overview","[{\"question\":\"Why is in-network machine learning inference effective for throughput and latency?\",\"answer\":\"Because inference runs within the network, data can move directly from source to destination without extra hops through inference servers or GPUs, reducing latency while maintaining high throughput.\"},{\"question\":\"What key challenges limit adoption of in-network ML in research?\",\"answer\":\"There is no general mapping methodology for translating diverse ML models to programmable data planes, and many solutions suffer from stage and memory explosion. Development also requires broad expertise across programmable networking, ML, and the target application domain.\"},{\"question\":\"How does Planter support rapid prototyping and portability?\",\"answer\":\"Planter provides end-to-end automated development, testing, and deployment, using three general mapping methodologies to enable a wide range of ML models across multiple architectures and target devices.\"}]","Planter - Rapid Prototyping of In-Network Machine Learning Inference | PDF",1785673766,45,{"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},"planter-rapid-prototyping-of-in-network-machine-learning-inference","",{"@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/planter-rapid-prototyping-of-in-network-machine-learning-inference/117104/",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 in-network machine learning inference effective for throughput and latency?","Question",{"text":75,"@type":76},"Because inference runs within the network, data can move directly from source to destination without extra hops through inference servers or GPUs, reducing latency while maintaining high throughput.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What key challenges limit adoption of in-network ML in research?",{"text":80,"@type":76},"There is no general mapping methodology for translating diverse ML models to programmable data planes, and many solutions suffer from stage and memory explosion. Development also requires broad expertise across programmable networking, ML, and the target application domain.",{"name":82,"@type":73,"acceptedAnswer":83},"How does Planter support rapid prototyping and portability?",{"text":84,"@type":76},"Planter provides end-to-end automated development, testing, and deployment, using three general mapping methodologies to enable a wide range of ML models across multiple architectures and target devices.","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"]