{"id":"https://openalex.org/W3177554119","doi":"https://doi.org/10.1109/tpds.2021.3052862","title":"MG-WFBP: Merging Gradients Wisely for Efficient Communication in Distributed Deep Learning","display_name":"MG-WFBP: Merging Gradients Wisely for Efficient Communication in Distributed Deep Learning","publication_year":2021,"publication_date":"2021-01-20","ids":{"openalex":"https://openalex.org/W3177554119","doi":"https://doi.org/10.1109/tpds.2021.3052862","mag":"3177554119"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2021.3052862","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2021.3052862","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-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/A5016836702","display_name":"Shaohuai Shi","orcid":"https://orcid.org/0000-0002-1418-5160"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":true,"raw_author_name":"Shaohuai Shi","raw_affiliation_strings":["Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Kowloon, Hong Kong, China"],"affiliations":[{"raw_affiliation_string":"Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Kowloon, Hong Kong, China","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100730785","display_name":"Xiaowen Chu","orcid":"https://orcid.org/0000-0001-9745-4372"},"institutions":[{"id":"https://openalex.org/I141568987","display_name":"Hong Kong Baptist University","ror":"https://ror.org/0145fw131","country_code":"HK","type":"education","lineage":["https://openalex.org/I141568987"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xiaowen Chu","raw_affiliation_strings":["Department of Computer Science, Hong Kong Baptist University, Kowloon, Hong Kong, China"],"affiliations":[{"raw_affiliation_string":"Department of Computer Science, Hong Kong Baptist University, Kowloon, Hong Kong, China","institution_ids":["https://openalex.org/I141568987"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5060171629","display_name":"Bo Li","orcid":"https://orcid.org/0000-0003-2955-750X"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Bo Li","raw_affiliation_strings":["Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Kowloon, Hong Kong, China"],"affiliations":[{"raw_affiliation_string":"Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Kowloon, Hong Kong, China","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5016836702"],"corresponding_institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"],"apc_list":null,"apc_paid":null,"fwci":2.4283,"has_fulltext":false,"cited_by_count":32,"citation_normalized_percentile":{"value":0.90761968,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":"32","issue":"8","first_page":"1903","last_page":"1917"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9993000030517578,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9993000030517578,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9988999962806702,"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"}},{"id":"https://openalex.org/T12676","display_name":"Machine Learning and ELM","score":0.9975000023841858,"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/computer-science","display_name":"Computer science","score":0.9083079099655151},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.7906200289726257},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.6814519166946411},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5598766803741455},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5332419872283936},{"id":"https://openalex.org/keywords/backpropagation","display_name":"Backpropagation","score":0.5228158235549927},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4953064024448395},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.4810248017311096},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.480008065700531},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.4274611473083496},{"id":"https://openalex.org/keywords/limit","display_name":"Limit (mathematics)","score":0.41279852390289307},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.405879408121109},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.39245712757110596},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.25562047958374023}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.9083079099655151},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7906200289726257},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.6814519166946411},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5598766803741455},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5332419872283936},{"id":"https://openalex.org/C155032097","wikidata":"https://www.wikidata.org/wiki/Q798503","display_name":"Backpropagation","level":3,"score":0.5228158235549927},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4953064024448395},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4810248017311096},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.480008065700531},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4274611473083496},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.41279852390289307},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.405879408121109},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.39245712757110596},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.25562047958374023},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/tpds.2021.3052862","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2021.3052862","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-article"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-108901","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-108901","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"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":"Article"},{"id":"pmh:oai:repository.ust.hk:1783.1-108901","is_oa":false,"landing_page_url":"http://gateway.isiknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcAuth=LinksAMR&SrcApp=PARTNER_APP&DestLinkType=FullRecord&DestApp=WOS&KeyUT=000622094200001","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"Article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":64,"referenced_works":["https://openalex.org/W1512808804","https://openalex.org/W1572016165","https://openalex.org/W1964981582","https://openalex.org/W2097117768","https://openalex.org/W2108598243","https://openalex.org/W2127941149","https://openalex.org/W2131613942","https://openalex.org/W2155893237","https://openalex.org/W2160054705","https://openalex.org/W2166221608","https://openalex.org/W2168231600","https://openalex.org/W2194775991","https://openalex.org/W2339765813","https://openalex.org/W2402144811","https://openalex.org/W2498764059","https://openalex.org/W2521757593","https://openalex.org/W2580688187","https://openalex.org/W2617766261","https://openalex.org/W2622263826","https://openalex.org/W2744387122","https://openalex.org/W2745269232","https://openalex.org/W2763421725","https://openalex.org/W2767393956","https://openalex.org/W2769644379","https://openalex.org/W2774000609","https://openalex.org/W2805245388","https://openalex.org/W2884711234","https://openalex.org/W2919632853","https://openalex.org/W2920397365","https://openalex.org/W2953384591","https://openalex.org/W2962758826","https://openalex.org/W2963112338","https://openalex.org/W2963339410","https://openalex.org/W2963446712","https://openalex.org/W2963674387","https://openalex.org/W2963786636","https://openalex.org/W2963803379","https://openalex.org/W2964004663","https://openalex.org/W2964324519","https://openalex.org/W2964350391","https://openalex.org/W2966527647","https://openalex.org/W2970971581","https://openalex.org/W2982475424","https://openalex.org/W2998689333","https://openalex.org/W3012220622","https://openalex.org/W3047357290","https://openalex.org/W3047537431","https://openalex.org/W3107997909","https://openalex.org/W4295312788","https://openalex.org/W6634263594","https://openalex.org/W6679154944","https://openalex.org/W6684859321","https://openalex.org/W6694260854","https://openalex.org/W6713134421","https://openalex.org/W6738460352","https://openalex.org/W6739622702","https://openalex.org/W6739693220","https://openalex.org/W6745245109","https://openalex.org/W6746200960","https://openalex.org/W6746839373","https://openalex.org/W6753209298","https://openalex.org/W6754886225","https://openalex.org/W6766978945","https://openalex.org/W6775649616"],"related_works":["https://openalex.org/W17155033","https://openalex.org/W3207760230","https://openalex.org/W1496222301","https://openalex.org/W1590307681","https://openalex.org/W4312814274","https://openalex.org/W4285370786","https://openalex.org/W2296488620","https://openalex.org/W2358353312","https://openalex.org/W2353836703","https://openalex.org/W4211135130"],"abstract_inverted_index":{"Distributed":[0],"synchronous":[1],"stochastic":[2],"gradient":[3],"descent":[4],"has":[5],"been":[6],"widely":[7],"used":[8],"to":[9,38,68,107,137,164,177,180],"train":[10],"deep":[11],"neural":[12],"networks":[13],"(DNNs)":[14],"on":[15,83],"computer":[16],"clusters.":[17],"With":[18],"the":[19,28,44,84,98,109,129,135,166,182],"increase":[20],"of":[21,59,66,168,175,186],"computational":[22],"power,":[23],"network":[24],"communications":[25,40,114],"generally":[26],"limit":[27],"system":[29],"scalability.":[30],"Wait-free":[31],"backpropagation":[32],"(WFBP)":[33],"is":[34],"a":[35,56,63,93,139],"popular":[36],"solution":[37,121,136],"overlap":[39],"with":[41,61],"computations":[42],"during":[43],"training":[45,110,130,141],"process.":[46],"In":[47],"this":[48],"article,":[49],"we":[50,102],"observe":[51],"that":[52,86,122,191],"many":[53],"DNNs":[54],"have":[55],"large":[57],"number":[58],"layers":[60],"only":[62],"small":[64],"amount":[65],"data":[67],"be":[69,124],"communicated":[70],"at":[71],"each":[72],"layer":[73],"in":[74,112,150,158],"distributed":[75,140],"training,":[76],"which":[77],"could":[78],"make":[79],"WFBP":[80,145],"inefficient.":[81],"Based":[82],"fact":[85],"merging":[87],"some":[88],"short":[89],"communication":[90,100],"tasks":[91],"into":[92],"single":[94],"one":[95],"can":[96,123],"reduce":[97],"overall":[99],"time,":[101],"formulate":[103],"an":[104,119],"optimization":[105],"problem":[106],"minimize":[108],"time":[111],"pipelining":[113],"and":[115,147,154],"computations.":[116],"We":[117,132,170],"derive":[118],"optimal":[120],"solved":[125],"efficiently":[126],"without":[127],"affecting":[128],"performance.":[131],"then":[133],"apply":[134],"propose":[138],"algorithm":[142],"named":[143],"merged-gradient":[144],"(MG-WFBP)":[146],"implement":[148],"it":[149],"two":[151],"platforms":[152],"Caffe":[153],"PyTorch.":[155],"Extensive":[156],"experiments":[157],"three":[159],"GPU":[160],"clusters":[161],"are":[162],"conducted":[163],"verify":[165],"effectiveness":[167],"MG-WFBP.":[169,187],"further":[171],"exploit":[172],"trace-based":[173],"simulations":[174],"4":[176],"2048":[178],"GPUs":[179],"explore":[181],"potential":[183],"scaling":[184,196],"efficiency":[185],"Experimental":[188],"results":[189],"show":[190],"MG-WFBP":[192],"achieves":[193],"much":[194],"better":[195],"performance":[197],"than":[198],"existing":[199],"methods.":[200]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":9},{"year":2022,"cited_by_count":6},{"year":2021,"cited_by_count":4}],"updated_date":"2026-04-02T15:55:50.835912","created_date":"2025-10-10T00:00:00"}