{"id":"https://openalex.org/W4318148006","doi":"https://doi.org/10.1109/bigdata55660.2022.10021133","title":"PaddleBox: Communication-Efficient TeraByte-Scale Model Training Framework for Online Advertising","display_name":"PaddleBox: Communication-Efficient TeraByte-Scale Model Training Framework for Online Advertising","publication_year":2022,"publication_date":"2022-12-17","ids":{"openalex":"https://openalex.org/W4318148006","doi":"https://doi.org/10.1109/bigdata55660.2022.10021133"},"language":"en","primary_location":{"id":"doi:10.1109/bigdata55660.2022.10021133","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/bigdata55660.2022.10021133","pdf_url":null,"source":{"id":"https://openalex.org/S4363607709","display_name":"2022 IEEE International Conference on Big Data (Big Data)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 IEEE International Conference on Big Data (Big Data)","raw_type":"proceedings-article"},"type":"article","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/A5075541712","display_name":"Weijie Zhao","orcid":"https://orcid.org/0000-0003-0967-1436"},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Weijie Zhao","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006865469","display_name":"Xuewu Jiao","orcid":"https://orcid.org/0009-0004-6530-4774"},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuewu Jiao","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100804855","display_name":"Mingqing Hu","orcid":null},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Mingqing Hu","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100781895","display_name":"Xiaoyun Li","orcid":"https://orcid.org/0000-0001-5730-2972"},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiaoyun Li","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107249133","display_name":"Xiangyu Zhang","orcid":"https://orcid.org/0000-0002-9544-2500"},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiangyu Zhang","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100435468","display_name":"Ping Li","orcid":"https://orcid.org/0000-0001-8272-6582"},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ping Li","raw_affiliation_strings":["Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","Cognitive Computing Lab, Baidu Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Baidu Research Baidu Search Ads (Phoenix Nest), Baidu Inc.,Cognitive Computing Lab,Washington,USA,98004","institution_ids":[]},{"raw_affiliation_string":"Cognitive Computing Lab, Baidu Inc., Beijing, China","institution_ids":["https://openalex.org/I98301712"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5075541712"],"corresponding_institution_ids":["https://openalex.org/I98301712"],"apc_list":null,"apc_paid":null,"fwci":0.4372,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.64138631,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":94},"biblio":{"volume":"12","issue":null,"first_page":"1401","last_page":"1408"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10203","display_name":"Recommender Systems and Techniques","score":0.9993000030517578,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10203","display_name":"Recommender Systems and Techniques","score":0.9993000030517578,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.9952999949455261,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T11165","display_name":"Image and Video Quality Assessment","score":0.994700014591217,"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/computer-science","display_name":"Computer science","score":0.8509776592254639},{"id":"https://openalex.org/keywords/terabyte","display_name":"Terabyte","score":0.7239781022071838},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.7147976160049438},{"id":"https://openalex.org/keywords/workflow","display_name":"Workflow","score":0.47159016132354736},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.4524768590927124},{"id":"https://openalex.org/keywords/click-through-rate","display_name":"Click-through rate","score":0.41991347074508667},{"id":"https://openalex.org/keywords/server","display_name":"Server","score":0.41359996795654297},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.40559059381484985},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3794325292110443},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.18569931387901306},{"id":"https://openalex.org/keywords/database","display_name":"Database","score":0.17477267980575562},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.15208080410957336},{"id":"https://openalex.org/keywords/information-retrieval","display_name":"Information retrieval","score":0.11896401643753052}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8509776592254639},{"id":"https://openalex.org/C199683683","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Terabyte","level":2,"score":0.7239781022071838},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7147976160049438},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.47159016132354736},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.4524768590927124},{"id":"https://openalex.org/C115174607","wikidata":"https://www.wikidata.org/wiki/Q1100934","display_name":"Click-through rate","level":2,"score":0.41991347074508667},{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.41359996795654297},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.40559059381484985},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3794325292110443},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.18569931387901306},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.17477267980575562},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.15208080410957336},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.11896401643753052},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/bigdata55660.2022.10021133","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/bigdata55660.2022.10021133","pdf_url":null,"source":{"id":"https://openalex.org/S4363607709","display_name":"2022 IEEE International Conference on Big Data (Big Data)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 IEEE International Conference on Big Data (Big Data)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/9","score":0.6200000047683716,"display_name":"Industry, innovation and infrastructure"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":48,"referenced_works":["https://openalex.org/W2045271686","https://openalex.org/W2060393849","https://openalex.org/W2064987260","https://openalex.org/W2083842231","https://openalex.org/W2096451472","https://openalex.org/W2146502635","https://openalex.org/W2156037541","https://openalex.org/W2339765813","https://openalex.org/W2475334473","https://openalex.org/W2512971201","https://openalex.org/W2517540742","https://openalex.org/W2548570154","https://openalex.org/W2560674852","https://openalex.org/W2604662567","https://openalex.org/W2605800822","https://openalex.org/W2612387305","https://openalex.org/W2723293840","https://openalex.org/W2751343396","https://openalex.org/W2793768763","https://openalex.org/W2886463271","https://openalex.org/W2950960796","https://openalex.org/W2978329087","https://openalex.org/W2984020950","https://openalex.org/W3080510735","https://openalex.org/W3094279629","https://openalex.org/W3104030692","https://openalex.org/W3122305203","https://openalex.org/W3154197656","https://openalex.org/W3173839890","https://openalex.org/W4245826835","https://openalex.org/W4289401659","https://openalex.org/W4293363567","https://openalex.org/W4320024053","https://openalex.org/W6628377381","https://openalex.org/W6631190155","https://openalex.org/W6638533276","https://openalex.org/W6679393576","https://openalex.org/W6681435938","https://openalex.org/W6728757088","https://openalex.org/W6747620207","https://openalex.org/W6753170497","https://openalex.org/W6753892653","https://openalex.org/W6762930437","https://openalex.org/W6773976177","https://openalex.org/W6774806506","https://openalex.org/W6785403911","https://openalex.org/W6800650132","https://openalex.org/W6838575114"],"related_works":["https://openalex.org/W1976914335","https://openalex.org/W2066858118","https://openalex.org/W2134017072","https://openalex.org/W2915208987","https://openalex.org/W1940452713","https://openalex.org/W2152256925","https://openalex.org/W1994777790","https://openalex.org/W2005567565","https://openalex.org/W2806040249","https://openalex.org/W848768768"],"abstract_inverted_index":{"Click-through":[0],"rate":[1,225],"(CTR)":[2],"prediction":[3,27],"is":[4,38,171,236],"one":[5],"of":[6,115,232,240,259],"the":[7,12,51,59,80,102,122,132,144,156,176,180,197,201,206,230,237,257],"most":[8],"crucial":[9],"components":[10],"in":[11,101,161,226,245],"online":[13],"advertising":[14],"industry.":[15],"In":[16,93,186],"order":[17],"to":[18,57,72,78,117,143,155,167,174],"produce":[19],"a":[20,31,49,113,191,214],"personalized":[21],"CTR":[22,26,247],"prediction,":[23],"an":[24],"industry-level":[25],"model":[28,52,216,248],"commonly":[29],"takes":[30],"high-dimensional":[32,60],"(\u223c":[33],"10<sup>12</sup>)":[34],"sparse":[35],"vector":[36],"(that":[37],"encoded":[39],"from":[40],"query":[41],"keywords,":[42],"user":[43],"portraits,":[44],"etc.)":[45],"as":[46,90],"input.":[47,61],"As":[48],"result,":[50],"requires":[53],"Terabyte":[54],"scale":[55],"parameters":[56],"embed":[58],"Hierarchical":[62],"distributed":[63,181],"GPU":[64,74,104],"parameter":[65],"server":[66],"has":[67],"been":[68],"developed":[69],"at":[70],"Baidu":[71],"enable":[73],"with":[75,128],"limited":[76],"memory":[77,87],"train":[79],"massive":[81],"network":[82],"by":[83],"leveraging":[84],"CPU":[85],"main":[86],"and":[88,111,138,221],"SSDs":[89],"secondary":[91],"storage.":[92],"this":[94,187,235],"work,":[95],"we":[96,189,212],"identify":[97],"two":[98],"major":[99],"challenges":[100],"existing":[103],"training":[105,193,262],"framework":[106],"for":[107,219],"massive-scale":[108],"ad":[109],"models":[110],"propose":[112,190],"collection":[114],"optimizations":[116],"tackle":[118],"these":[119],"challenges:":[120],"(a)":[121],"GPU,":[123],"CPU,":[124],"SSD":[125],"rapidly":[126],"communicate":[127],"each":[129],"other":[130],"during":[131],"training.":[133,249],"The":[134,147],"connections":[135],"between":[136,209],"GPUs":[137,160],"CPUs":[139],"are":[140],"non-uniform":[141],"due":[142],"hardware":[145,157,198],"topology.":[146],"data":[148,255],"communication":[149,208],"route":[150],"should":[151],"be":[152],"optimized":[153],"according":[154],"topology;":[158],"(b)":[159],"different":[162],"computing":[163,210],"nodes":[164],"frequently":[165],"communicates":[166],"synchronize":[168],"parameters.":[169],"It":[170],"thus":[172],"required":[173],"optimize":[175],"communications":[177],"so":[178],"that":[179,195],"system":[182],"can":[183],"become":[184],"scalable.":[185],"paper,":[188],"hardware-aware":[192],"workflow":[194],"couples":[196],"topology":[199],"into":[200],"algorithm":[202,218],"design.":[203],"To":[204,229],"reduce":[205],"extensive":[207],"nodes,":[211],"introduce":[213],"k-step":[215,241],"merging":[217],"Adam":[220],"provide":[222],"its":[223],"convergence":[224],"non-convex":[227],"optimization.":[228],"best":[231],"our":[233,260],"knowledge,":[234],"first":[238],"application":[239],"adaptive":[242],"optimization":[243],"method":[244],"industrial":[246],"Experiments":[250],"on":[251],"commercial":[252],"search":[253],"ads":[254],"confirm":[256],"effectiveness":[258],"proposed":[261],"framework.":[263]},"counts_by_year":[{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":1},{"year":2022,"cited_by_count":1}],"updated_date":"2025-12-23T23:11:35.936235","created_date":"2025-10-10T00:00:00"}