{"id":"https://openalex.org/W3081146346","doi":"https://doi.org/10.1145/3394486.3403342","title":"Large-Scale Training System for 100-Million Classification at Alibaba","display_name":"Large-Scale Training System for 100-Million Classification at Alibaba","publication_year":2020,"publication_date":"2020-08-20","ids":{"openalex":"https://openalex.org/W3081146346","doi":"https://doi.org/10.1145/3394486.3403342","mag":"3081146346"},"language":"en","primary_location":{"id":"doi:10.1145/3394486.3403342","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3394486.3403342","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2102.06025","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Liuyihan Song","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liuyihan Song","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Pan Pan","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Pan Pan","raw_affiliation_strings":["Alibaba Group, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Beijing, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Kang Zhao","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kang Zhao","raw_affiliation_strings":["Alibaba Group, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Beijing, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Hao Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hao Yang","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yiming Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yiming Chen","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yingya Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yingya Zhang","raw_affiliation_strings":["Alibaba Group, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Beijing, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yinghui Xu","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yinghui Xu","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"last","author":{"id":null,"display_name":"Rong Jin","orcid":null},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Rong Jin","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I45928872"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":18,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2909","last_page":"2930"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11775","display_name":"COVID-19 diagnosis using AI","score":0.9742000102996826,"subfield":{"id":"https://openalex.org/subfields/2741","display_name":"Radiology, Nuclear Medicine and Imaging"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},"topics":[{"id":"https://openalex.org/T11775","display_name":"COVID-19 diagnosis using AI","score":0.9742000102996826,"subfield":{"id":"https://openalex.org/subfields/2741","display_name":"Radiology, Nuclear Medicine and Imaging"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9684000015258789,"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"}},{"id":"https://openalex.org/T12072","display_name":"Machine Learning and Algorithms","score":0.9580000042915344,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/softmax-function","display_name":"Softmax function","score":0.9192000031471252},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.6136000156402588},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5532000064849854},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5465999841690063},{"id":"https://openalex.org/keywords/classifier","display_name":"Classifier (UML)","score":0.5163999795913696},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4966000020503998},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.49480000138282776},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4499000012874603}],"concepts":[{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.9192000031471252},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8274999856948853},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.629800021648407},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6136000156402588},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5532000064849854},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5465999841690063},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.5163999795913696},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4966000020503998},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4957999885082245},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.49480000138282776},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4499000012874603},{"id":"https://openalex.org/C2776857766","wikidata":"https://www.wikidata.org/wiki/Q7832987","display_name":"Training system","level":2,"score":0.37299999594688416},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.3102000057697296},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3082999885082245},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3000999987125397},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.2745000123977661},{"id":"https://openalex.org/C90673727","wikidata":"https://www.wikidata.org/wiki/Q901718","display_name":"Product (mathematics)","level":2,"score":0.27239999175071716},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.2660999894142151},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.26499998569488525}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3394486.3403342","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3394486.3403342","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2102.06025","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2102.06025","pdf_url":"https://arxiv.org/pdf/2102.06025","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2102.06025","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2102.06025","pdf_url":"https://arxiv.org/pdf/2102.06025","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":10,"referenced_works":["https://openalex.org/W2100714283","https://openalex.org/W2194775991","https://openalex.org/W2606891064","https://openalex.org/W2744136723","https://openalex.org/W2962747323","https://openalex.org/W2963433607","https://openalex.org/W2963932686","https://openalex.org/W2964369530","https://openalex.org/W2969985801","https://openalex.org/W2986277806"],"related_works":[],"abstract_inverted_index":{"In":[0,57],"the":[1,47,53,80,97,106,113,150,154,207],"last":[2,54],"decades,":[3],"extreme":[4],"classification":[5],"has":[6,15],"become":[7],"an":[8,178],"essential":[9],"topic":[10],"for":[11],"deep":[12,39],"learning.":[13],"It":[14],"achieved":[16],"great":[17],"success":[18],"in":[19,23,52,197],"many":[20],"areas,":[21],"especially":[22],"computer":[24],"vision":[25],"and":[26,49,101,104,122,145,165],"natural":[27],"language":[28],"processing":[29],"(NLP).":[30],"However,":[31],"it":[32],"is":[33],"very":[34],"challenging":[35],"to":[36,46,66,78,111,135],"train":[37,185],"a":[38,62,73,87,118,123,130,186,203],"model":[40,147],"with":[41,206],"millions":[42],"of":[43,108,152,161,169,188],"classes":[44,191],"due":[45],"memory":[48,99],"computation":[50,102],"explosion":[51],"output":[55],"layer.":[56],"this":[58],"paper,":[59],"we":[60,71,85,116,128,157,183],"propose":[61,86,117],"large-scale":[63],"training":[64,76,81,138,163,170,210],"system":[65,164],"address":[67],"these":[68],"challenges.":[69],"First,":[70],"build":[72],"hybrid":[74],"parallel":[75],"framework":[77],"make":[79],"process":[82],"feasible.":[83],"Second,":[84],"novel":[88],"softmax":[89,209],"variation":[90],"named":[91],"KNN":[92],"softmax,":[93],"which":[94],"reduces":[95],"both":[96],"GPU":[98],"consumption":[100],"costs":[103],"improves":[105],"throughput":[107,160],"training.":[109],"Then,":[110],"eliminate":[112],"communication":[114],"overhead,":[115],"new":[119],"overlapping":[120],"pipeline":[121],"gradient":[124],"sparsification":[125],"method.":[126],"Furthermore,":[127],"design":[129],"fast":[131],"continuous":[132],"convergence":[133],"strategy":[134],"reduce":[136,166],"total":[137],"iterations":[139],"by":[140],"adaptively":[141],"adjusting":[142],"learning":[143],"rate":[144],"updating":[146],"parameters.":[148],"With":[149],"help":[151],"all":[153],"proposed":[155],"methods,":[156],"gain":[158],"3.9\u00d7":[159],"our":[162],"almost":[167],"60%":[168],"iterations.":[171],"The":[172],"experimental":[173],"results":[174],"show":[175],"that":[176],"using":[177],"in-house":[179],"256":[180],"GPUs":[181],"cluster,":[182],"could":[184],"classifier":[187],"100":[189],"million":[190],"on":[192],"Alibaba":[193],"Retail":[194],"Product":[195],"Dataset":[196],"about":[198],"five":[199],"days":[200],"while":[201],"achieving":[202],"comparable":[204],"accuracy":[205],"naive":[208],"process.":[211]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":4},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":5},{"year":2020,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2020-09-01T00:00:00"}