{"id":"https://openalex.org/W4410049379","doi":"https://doi.org/10.1145/3680256.3721266","title":"IPA: Inference Pipeline Adaptation to Achieve High Accuracy and Cost-Efficiency","display_name":"IPA: Inference Pipeline Adaptation to Achieve High Accuracy and Cost-Efficiency","publication_year":2025,"publication_date":"2025-05-03","ids":{"openalex":"https://openalex.org/W4410049379","doi":"https://doi.org/10.1145/3680256.3721266"},"language":"en","primary_location":{"id":"doi:10.1145/3680256.3721266","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3680256.3721266","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Companion of the 16th ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5019689955","display_name":"Saeid Ghafouri","orcid":"https://orcid.org/0000-0003-3799-5702"},"institutions":[{"id":"https://openalex.org/I126231945","display_name":"Queen's University Belfast","ror":"https://ror.org/00hswnk62","country_code":"GB","type":"education","lineage":["https://openalex.org/I126231945"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Saeid Ghafouri","raw_affiliation_strings":["Queen's University Belfast, Belfast, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-3799-5702","affiliations":[{"raw_affiliation_string":"Queen's University Belfast, Belfast, United Kingdom","institution_ids":["https://openalex.org/I126231945"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045974365","display_name":"Kamran Razavi","orcid":"https://orcid.org/0000-0002-3232-5657"},"institutions":[{"id":"https://openalex.org/I31512782","display_name":"Technische Universit\u00e4t Darmstadt","ror":"https://ror.org/05n911h24","country_code":"DE","type":"education","lineage":["https://openalex.org/I31512782"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Kamran Razavi","raw_affiliation_strings":["Technical University of Darmstadt, Darmstadt, Germany"],"raw_orcid":"https://orcid.org/0000-0002-3232-5657","affiliations":[{"raw_affiliation_string":"Technical University of Darmstadt, Darmstadt, Germany","institution_ids":["https://openalex.org/I31512782"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091376011","display_name":"Mehran Salmani","orcid":"https://orcid.org/0009-0008-6362-0967"},"institutions":[{"id":"https://openalex.org/I119449181","display_name":"Technische Universit\u00e4t Ilmenau","ror":"https://ror.org/01weqhp73","country_code":"DE","type":"education","lineage":["https://openalex.org/I119449181"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Mehran Salmani","raw_affiliation_strings":["Technical University of Ilmenau, Ilmenau, Germany"],"raw_orcid":"https://orcid.org/0009-0008-6362-0967","affiliations":[{"raw_affiliation_string":"Technical University of Ilmenau, Ilmenau, Germany","institution_ids":["https://openalex.org/I119449181"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Alireza Sanaee","orcid":"https://orcid.org/0000-0001-6461-1650"},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Alireza Sanaee","raw_affiliation_strings":["University of Cambridge, Cambridge, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0001-6461-1650","affiliations":[{"raw_affiliation_string":"University of Cambridge, Cambridge, United Kingdom","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008852667","display_name":"Tania Lorido Botran","orcid":"https://orcid.org/0009-0002-8836-3983"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tania Lorido Botran","raw_affiliation_strings":["Roblox Corporation, San Jose, USA"],"raw_orcid":"https://orcid.org/0009-0002-8836-3983","affiliations":[{"raw_affiliation_string":"Roblox Corporation, San Jose, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100403099","display_name":"Lin Wang","orcid":"https://orcid.org/0000-0001-7181-6128"},"institutions":[{"id":"https://openalex.org/I206945453","display_name":"Paderborn University","ror":"https://ror.org/058kzsd48","country_code":"DE","type":"education","lineage":["https://openalex.org/I206945453"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Lin Wang","raw_affiliation_strings":["Paderborn University, Paderborn, Germany"],"raw_orcid":"https://orcid.org/0000-0001-7181-6128","affiliations":[{"raw_affiliation_string":"Paderborn University, Paderborn, Germany","institution_ids":["https://openalex.org/I206945453"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002561390","display_name":"Joseph Doyle","orcid":"https://orcid.org/0000-0003-1840-9616"},"institutions":[{"id":"https://openalex.org/I166337079","display_name":"Queen Mary University of London","ror":"https://ror.org/026zzn846","country_code":"GB","type":"education","lineage":["https://openalex.org/I124357947","https://openalex.org/I166337079"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Joseph Doyle","raw_affiliation_strings":["Queen Mary University of London, London, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-1840-9616","affiliations":[{"raw_affiliation_string":"Queen Mary University of London, London, United Kingdom","institution_ids":["https://openalex.org/I166337079"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5064540250","display_name":"Pooyan Jamshidi","orcid":"https://orcid.org/0000-0002-9342-0703"},"institutions":[{"id":"https://openalex.org/I155781252","display_name":"University of South Carolina","ror":"https://ror.org/02b6qw903","country_code":"US","type":"education","lineage":["https://openalex.org/I155781252"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Pooyan Jamshidi","raw_affiliation_strings":["University of South Carolina, Columbia, South Carolina, USA"],"raw_orcid":"https://orcid.org/0000-0002-9342-0703","affiliations":[{"raw_affiliation_string":"University of South Carolina, Columbia, South Carolina, USA","institution_ids":["https://openalex.org/I155781252"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":7,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2","last_page":"3"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11220","display_name":"Water Systems and Optimization","score":0.8935999870300293,"subfield":{"id":"https://openalex.org/subfields/2205","display_name":"Civil and Structural Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11220","display_name":"Water Systems and Optimization","score":0.8935999870300293,"subfield":{"id":"https://openalex.org/subfields/2205","display_name":"Civil and Structural Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12169","display_name":"Non-Destructive Testing Techniques","score":0.843999981880188,"subfield":{"id":"https://openalex.org/subfields/2210","display_name":"Mechanical Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13050","display_name":"Oil and Gas Production Techniques","score":0.7578999996185303,"subfield":{"id":"https://openalex.org/subfields/2212","display_name":"Ocean Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7645694017410278},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.7607653737068176},{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.7290117740631104},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7178621292114258},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3417019844055176},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.09772452712059021},{"id":"https://openalex.org/keywords/psychology","display_name":"Psychology","score":0.05111017823219299}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7645694017410278},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.7607653737068176},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.7290117740631104},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7178621292114258},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3417019844055176},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.09772452712059021},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.05111017823219299},{"id":"https://openalex.org/C169760540","wikidata":"https://www.wikidata.org/wiki/Q207011","display_name":"Neuroscience","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3680256.3721266","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3680256.3721266","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Companion of the 16th ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"},{"id":"pmh:oai:tubiblio.ulb.tu-darmstadt.de:155189","is_oa":false,"landing_page_url":"http://tubiblio.ulb.tu-darmstadt.de/view/person/Ghafouri=3ASaeid=3A=3A.html>","pdf_url":null,"source":{"id":"https://openalex.org/S4377196390","display_name":"TUbilio (Technical University of Darmstadt)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I31512782","host_organization_name":"Technische Universit\u00e4t Darmstadt","host_organization_lineage":["https://openalex.org/I31512782"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"Konferenzver\u00f6ffentlichung"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.5099999904632568}],"awards":[{"id":"https://openalex.org/G1036627719","display_name":null,"funder_award_id":"2007202, 2107463, 2038080, and 2233873","funder_id":"https://openalex.org/F4320323817","funder_display_name":"Universitas Brawijaya"}],"funders":[{"id":"https://openalex.org/F4320323817","display_name":"Universitas Brawijaya","ror":"https://ror.org/01wk3d929"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":1,"referenced_works":["https://openalex.org/W4395481220"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Efficiently":[0],"optimizing":[1,79],"multi-model":[2],"inference":[3,10,29,75,158,214],"pipelines":[4,76,159],"for":[5],"fast,":[6],"accurate,":[7],"and":[8,39,48,93,115,125,171],"cost-effective":[9],"is":[11,87,129,219],"a":[12,96,152,187,204],"critical":[13],"challenge":[14],"in":[15,95,132,136,151,198],"machine":[16],"learning":[17,28],"production":[18],"systems.":[19],"Our":[20],"work":[21,102],"introduces":[22],"IPA,":[23],"an":[24],"innovative":[25],"online":[26],"deep":[27],"pipeline":[30,68,98,145],"adaptation":[31],"system.":[32],"IPA":[33,59,109,178],"dynamically":[34],"configures":[35],"batch":[36],"sizes,":[37],"replication,":[38],"model":[40,137],"variants":[41,138],"to":[42,84,118,142,184,207],"simultaneously":[43],"optimize":[44],"accuracy,":[45,123],"minimize":[46],"costs,":[47],"meet":[49],"user-defined":[50],"latency":[51],"Service":[52],"Level":[53],"Agreements":[54],"(SLAs).":[55],"Unlike":[56],"existing":[57],"approaches,":[58],"captures":[60],"the":[61,208],"correlation":[62],"between":[63,122],"configuration":[64],"changes":[65],"across":[66],"multiple":[67],"stages,":[69],"enabling":[70],"holistic":[71],"end-to-end":[72],"optimization":[73,90,209],"of":[74,91,191,210],"rather":[77],"than":[78,193],"each":[80],"stage":[81],"individually.":[82],"Key":[83],"IPA's":[85],"innovation":[86],"its":[88],"joint":[89],"accuracy":[92,181],"cost":[94,189],"multi-stage":[97,212],"setting.":[99],"Although":[100],"previous":[101],"has":[103],"focused":[104],"predominantly":[105],"on":[106],"cost-aware":[107],"optimizations,":[108],"leverages":[110],"accuracy-aware":[111],"adaptations,":[112],"combining":[113],"autoscaling":[114],"model-switching":[116],"techniques":[117],"achieve":[119],"precise":[120],"trade-offs":[121,201],"cost,":[124],"latency.":[126],"This":[127],"approach":[128],"particularly":[130],"advantageous":[131],"scenarios":[133],"where":[134],"heterogeneity":[135],"can":[139],"be":[140],"exploited":[141],"enhance":[143],"overall":[144],"performance":[146],"dynamically.":[147],"Through":[148],"extensive":[149],"evaluations":[150],"Kubernetes":[153],"implementation":[154],"with":[155,186],"five":[156],"real-world":[157],"(Video":[160],"Monitoring,":[161],"Audio":[162,165],"Question":[163,169],"Answering,":[164,170],"Sentiment":[166],"Analysis,":[167],"Summarisation":[168],"Natural":[172],"Language":[173],"Processing),":[174],"we":[175],"demonstrate":[176],"that":[177],"improves":[179],"normalized":[180],"by":[182],"up":[183],"35%":[185],"negligible":[188],"increase":[190],"less":[192],"5%.":[194],"The":[195,216],"system's":[196],"effectiveness":[197],"achieving":[199],"granular":[200],"makes":[202],"it":[203],"valuable":[205],"contribution":[206],"complex,":[211],"ML":[213],"pipelines.":[215],"replication":[217],"package":[218],"available":[220],"at":[221],"https://github.com/reconfigurable-ml-pipeline/ipa.":[222]},"counts_by_year":[{"year":2025,"cited_by_count":2}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}