{"id":"https://openalex.org/W4400433711","doi":"https://doi.org/10.1145/3625549.3658688","title":"Loki: A System for Serving ML Inference Pipelines with Hardware and Accuracy Scaling","display_name":"Loki: A System for Serving ML Inference Pipelines with Hardware and Accuracy Scaling","publication_year":2024,"publication_date":"2024-06-03","ids":{"openalex":"https://openalex.org/W4400433711","doi":"https://doi.org/10.1145/3625549.3658688"},"language":"en","primary_location":{"id":"doi:10.1145/3625549.3658688","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3625549.3658688","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3625549.3658688?download=true","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd International Symposium on High-Performance Parallel and Distributed Computing","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3625549.3658688?download=true","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5103133518","display_name":"Sohaib Ahmad","orcid":"https://orcid.org/0009-0008-3892-9460"},"institutions":[{"id":"https://openalex.org/I24603500","display_name":"University of Massachusetts Amherst","ror":"https://ror.org/0072zz521","country_code":"US","type":"education","lineage":["https://openalex.org/I24603500"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sohaib Ahmad","raw_affiliation_strings":["University of Massachusetts Amherst, Amherst, Massachusetts, United States of America"],"raw_orcid":"https://orcid.org/0009-0008-3892-9460","affiliations":[{"raw_affiliation_string":"University of Massachusetts Amherst, Amherst, Massachusetts, United States of America","institution_ids":["https://openalex.org/I24603500"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085489377","display_name":"Hui Guan","orcid":"https://orcid.org/0000-0001-9128-2231"},"institutions":[{"id":"https://openalex.org/I24603500","display_name":"University of Massachusetts Amherst","ror":"https://ror.org/0072zz521","country_code":"US","type":"education","lineage":["https://openalex.org/I24603500"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hui Guan","raw_affiliation_strings":["University of Massachusetts Amherst, Amherst, Massachusetts, United States of America"],"raw_orcid":"https://orcid.org/0000-0001-9128-2231","affiliations":[{"raw_affiliation_string":"University of Massachusetts Amherst, Amherst, Massachusetts, United States of America","institution_ids":["https://openalex.org/I24603500"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037538509","display_name":"Ramesh K. Sitaraman","orcid":"https://orcid.org/0000-0003-0558-6875"},"institutions":[{"id":"https://openalex.org/I24603500","display_name":"University of Massachusetts Amherst","ror":"https://ror.org/0072zz521","country_code":"US","type":"education","lineage":["https://openalex.org/I24603500"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ramesh K. Sitaraman","raw_affiliation_strings":["University of Massachusetts Amherst, Amherst, Massachusetts, United States of America"],"raw_orcid":"https://orcid.org/0000-0003-0558-6875","affiliations":[{"raw_affiliation_string":"University of Massachusetts Amherst, Amherst, Massachusetts, United States of America","institution_ids":["https://openalex.org/I24603500"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":3.0276,"has_fulltext":true,"cited_by_count":7,"citation_normalized_percentile":{"value":0.92440417,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"267","last_page":"280"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9980999827384949,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9980999827384949,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9943000078201294,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.989799976348877,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.7136930227279663},{"id":"https://openalex.org/keywords/pipeline-transport","display_name":"Pipeline transport","score":0.7052876949310303},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6638679504394531},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6059097051620483},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.3497292399406433},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.20073390007019043},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.17427560687065125},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.09693130850791931}],"concepts":[{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.7136930227279663},{"id":"https://openalex.org/C175309249","wikidata":"https://www.wikidata.org/wiki/Q725864","display_name":"Pipeline transport","level":2,"score":0.7052876949310303},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6638679504394531},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6059097051620483},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.3497292399406433},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.20073390007019043},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.17427560687065125},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.09693130850791931},{"id":"https://openalex.org/C87717796","wikidata":"https://www.wikidata.org/wiki/Q146326","display_name":"Environmental engineering","level":1,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3625549.3658688","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3625549.3658688","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3625549.3658688?download=true","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd International Symposium on High-Performance Parallel and Distributed Computing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2407.03583","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2407.03583","pdf_url":"https://arxiv.org/pdf/2407.03583","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.1145/3625549.3658688","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3625549.3658688","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3625549.3658688?download=true","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd International Symposium on High-Performance Parallel and Distributed Computing","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Decent work and economic growth","score":0.6399999856948853,"id":"https://metadata.un.org/sdg/8"}],"awards":[{"id":"https://openalex.org/G1226108236","display_name":null,"funder_award_id":"DMS-2220211","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G13649008","display_name":null,"funder_award_id":"CNS-2224054","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G1441593048","display_name":"CAREER: Adaptive Deep Learning Systems Towards Edge Intelligence","funder_award_id":"2338512","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G183063246","display_name":null,"funder_award_id":"CNS-1901137","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G1957757468","display_name":"Collaborative Research: CSR: Medium: MemDrive: Memory-Driven Full-Stack Collaboration for Autonomous Embedded Systems","funder_award_id":"2312396","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G2838097271","display_name":null,"funder_award_id":"CNS-2106463","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G3686290445","display_name":"Collaborative Research: CNS Core: Medium: miVirtualSeat: Semantics-aware Content Distribution for Immersive Meeting Environments","funder_award_id":"2106463","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G3839821294","display_name":null,"funder_award_id":"CNS-2338512","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G4072858779","display_name":null,"funder_award_id":"CNS-1763617","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G4672264128","display_name":null,"funder_award_id":"CNS-2312396","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G6976256797","display_name":"CSR: Medium: Collaborative Research: Foundations of Cache Network Operations for Content Delivery","funder_award_id":"1763617","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7264624474","display_name":"CNS Core: Small: Transparently Scaling Graph Neural Network Training to Large-Scale Models and Graphs","funder_award_id":"2224054","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8565613624","display_name":"CNS Core: Medium: Collaborative Research: Scalable Dissemination and Navigation of Video 360 Content for Personalized Viewing","funder_award_id":"1901137","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320309480","display_name":"Nvidia","ror":"https://ror.org/03jdj4y14"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4400433711.pdf","grobid_xml":"https://content.openalex.org/works/W4400433711.grobid-xml"},"referenced_works_count":24,"referenced_works":["https://openalex.org/W95608104","https://openalex.org/W2034466462","https://openalex.org/W2194775991","https://openalex.org/W2604319603","https://openalex.org/W2734941459","https://openalex.org/W2772948367","https://openalex.org/W2897268228","https://openalex.org/W2941938531","https://openalex.org/W2962746093","https://openalex.org/W2982157693","https://openalex.org/W3095488153","https://openalex.org/W3209166877","https://openalex.org/W3210617645","https://openalex.org/W3210776666","https://openalex.org/W4214690606","https://openalex.org/W4254052724","https://openalex.org/W4282577879","https://openalex.org/W4313229743","https://openalex.org/W4366564134","https://openalex.org/W4368755837","https://openalex.org/W4372263604","https://openalex.org/W4388874804","https://openalex.org/W4394892775","https://openalex.org/W6893711219"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052","https://openalex.org/W2382290278","https://openalex.org/W4395014643"],"abstract_inverted_index":{"The":[0],"rapid":[1],"adoption":[2],"of":[3,11,72,171],"machine":[4],"learning":[5],"(ML)":[6],"has":[7],"underscored":[8],"the":[9,53,70,114,168],"importance":[10],"serving":[12,124],"ML":[13,73,87],"models":[14,74,88],"with":[15,101,128,191,208],"high":[16],"throughput":[17,216],"and":[18,89,131,144,155,213],"resource":[19,107,142],"efficiency.":[20],"Traditional":[21],"approaches":[22],"to":[23,50,75,91,105,183,198],"managing":[24,94],"increasing":[25,36],"query":[26,78,147],"demands":[27],"have":[28],"predominantly":[29],"focused":[30],"on":[31,186,211],"hardware":[32,102,130,187],"scaling,":[33,103,167],"which":[34],"involves":[35],"server":[37],"count":[38],"or":[39,56],"computing":[40],"power.":[41],"However,":[42],"this":[43,116],"strategy":[44],"can":[45,175],"often":[46],"be":[47,176],"impractical":[48],"due":[49],"limitations":[51],"in":[52,202],"available":[54],"budget":[55],"compute":[57],"resources.":[58],"As":[59],"an":[60,136,145],"alternative,":[61],"accuracy":[62,71,82,132,154,166,212],"scaling":[63,83],"offers":[64],"a":[65,120,172,199],"promising":[66],"solution":[67],"by":[68,178],"adjusting":[69],"accommodate":[76],"fluctuating":[77],"demands.":[79,217],"Yet,":[80],"existing":[81],"techniques":[84],"target":[85],"independent":[86],"tend":[90],"underperform":[92],"while":[93,214],"inference":[95,125],"pipelines.":[96],"Furthermore,":[97],"they":[98],"lack":[99],"integration":[100],"leading":[104],"potential":[106],"inefficiencies":[108],"during":[109],"low-demand":[110],"periods.":[111],"To":[112],"address":[113],"limitations,":[115],"paper":[117],"introduces":[118],"Loki,":[119],"system":[121,153],"designed":[122],"for":[123,140],"pipelines":[126],"effectively":[127],"both":[129],"scaling.":[133,188],"Loki":[134,195],"incorporates":[135],"innovative":[137],"theoretical":[138],"framework":[139],"optimal":[141],"allocation":[143],"effective":[146,169],"routing":[148],"algorithm,":[149],"aimed":[150],"at":[151],"improving":[152],"minimizing":[156],"latency":[157],"deadline":[158],"violations.":[159],"Our":[160],"empirical":[161],"evaluation":[162],"demonstrates":[163],"that":[164],"through":[165],"capacity":[170],"fixed-size":[173],"cluster":[174],"enhanced":[177],"more":[179],"than":[180],"2.7\u00d7":[181],"compared":[182,190],"relying":[184],"solely":[185],"When":[189],"state-of-the-art":[192],"inference-serving":[193],"systems,":[194],"achieves":[196],"up":[197],"10\u00d7":[200],"reduction":[201],"Service":[203],"Level":[204],"Objective":[205],"(SLO)":[206],"violations,":[207],"minimal":[209],"compromises":[210],"fulfilling":[215]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":1}],"updated_date":"2026-06-22T08:00:12.763002","created_date":"2025-10-10T00:00:00"}