{"id":"https://openalex.org/W3136347532","doi":"https://doi.org/10.1109/bigdata50022.2020.9378178","title":"Communication-Efficient Local Stochastic Gradient Descent for Scalable Deep Learning","display_name":"Communication-Efficient Local Stochastic Gradient Descent for Scalable Deep Learning","publication_year":2020,"publication_date":"2020-12-10","ids":{"openalex":"https://openalex.org/W3136347532","doi":"https://doi.org/10.1109/bigdata50022.2020.9378178","mag":"3136347532"},"language":"en","primary_location":{"id":"doi:10.1109/bigdata50022.2020.9378178","is_oa":false,"landing_page_url":"https://doi.org/10.1109/bigdata50022.2020.9378178","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2020 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/A5100698554","display_name":"Sunwoo Lee","orcid":"https://orcid.org/0000-0001-6334-3068"},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Sunwoo Lee","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA","institution_ids":["https://openalex.org/I111979921"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013922013","display_name":"Qiao Kang","orcid":"https://orcid.org/0000-0001-8552-6551"},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Qiao Kang","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA","institution_ids":["https://openalex.org/I111979921"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004659592","display_name":"Ankit Agrawal","orcid":"https://orcid.org/0000-0002-5519-0302"},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ankit Agrawal","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA","institution_ids":["https://openalex.org/I111979921"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074976770","display_name":"Alok Choudhary","orcid":"https://orcid.org/0000-0001-8152-6319"},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Alok Choudhary","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA","institution_ids":["https://openalex.org/I111979921"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5047602285","display_name":"Wei\u2010keng Liao","orcid":"https://orcid.org/0009-0008-9411-2543"},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wei-keng Liao","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Northwestern University, Evanston, USA","institution_ids":["https://openalex.org/I111979921"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5100698554"],"corresponding_institution_ids":["https://openalex.org/I111979921"],"apc_list":null,"apc_paid":null,"fwci":0.2743,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.66945356,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"718","last_page":"727"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9998999834060669,"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"}},"topics":[{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9998999834060669,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9997000098228455,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.9972000122070312,"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/scalability","display_name":"Scalability","score":0.7967511415481567},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7835606336593628},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.7771989703178406},{"id":"https://openalex.org/keywords/convergence","display_name":"Convergence (economics)","score":0.5325295925140381},{"id":"https://openalex.org/keywords/local-optimum","display_name":"Local optimum","score":0.5316871404647827},{"id":"https://openalex.org/keywords/parallelism","display_name":"Parallelism (grammar)","score":0.5036205649375916},{"id":"https://openalex.org/keywords/data-parallelism","display_name":"Data parallelism","score":0.48641228675842285},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.46029505133628845},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.4493074417114258},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.4266554117202759},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4262089431285858},{"id":"https://openalex.org/keywords/train","display_name":"Train","score":0.42429155111312866},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.4220725893974304},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3900406062602997},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.348388135433197},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.2912384867668152},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.22381463646888733}],"concepts":[{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7967511415481567},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7835606336593628},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.7771989703178406},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.5325295925140381},{"id":"https://openalex.org/C141934464","wikidata":"https://www.wikidata.org/wiki/Q3305386","display_name":"Local optimum","level":2,"score":0.5316871404647827},{"id":"https://openalex.org/C2781172179","wikidata":"https://www.wikidata.org/wiki/Q853109","display_name":"Parallelism (grammar)","level":2,"score":0.5036205649375916},{"id":"https://openalex.org/C61483411","wikidata":"https://www.wikidata.org/wiki/Q3124522","display_name":"Data parallelism","level":3,"score":0.48641228675842285},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.46029505133628845},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4493074417114258},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.4266554117202759},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4262089431285858},{"id":"https://openalex.org/C190839683","wikidata":"https://www.wikidata.org/wiki/Q2448197","display_name":"Train","level":2,"score":0.42429155111312866},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.4220725893974304},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3900406062602997},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.348388135433197},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2912384867668152},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.22381463646888733},{"id":"https://openalex.org/C58640448","wikidata":"https://www.wikidata.org/wiki/Q42515","display_name":"Cartography","level":1,"score":0.0},{"id":"https://openalex.org/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"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/C50522688","wikidata":"https://www.wikidata.org/wiki/Q189833","display_name":"Economic growth","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}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/bigdata50022.2020.9378178","is_oa":false,"landing_page_url":"https://doi.org/10.1109/bigdata50022.2020.9378178","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2020 IEEE International Conference on Big Data (Big Data)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320306084","display_name":"U.S. Department of Energy","ror":"https://ror.org/01bj3aw27"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":37,"referenced_works":["https://openalex.org/W1522301498","https://openalex.org/W2059300917","https://openalex.org/W2105549957","https://openalex.org/W2131613942","https://openalex.org/W2194775991","https://openalex.org/W2263490141","https://openalex.org/W2617242334","https://openalex.org/W2622263826","https://openalex.org/W2739757502","https://openalex.org/W2751713192","https://openalex.org/W2787299084","https://openalex.org/W2900182564","https://openalex.org/W2904556356","https://openalex.org/W2911342633","https://openalex.org/W2963372104","https://openalex.org/W2963655672","https://openalex.org/W2963702144","https://openalex.org/W2963959597","https://openalex.org/W2964121744","https://openalex.org/W2964243509","https://openalex.org/W2970020383","https://openalex.org/W2972163183","https://openalex.org/W3037047862","https://openalex.org/W3037875189","https://openalex.org/W3118608800","https://openalex.org/W3124352525","https://openalex.org/W4297781872","https://openalex.org/W6631190155","https://openalex.org/W6692956712","https://openalex.org/W6726983090","https://openalex.org/W6738534199","https://openalex.org/W6739622702","https://openalex.org/W6754818105","https://openalex.org/W6755251367","https://openalex.org/W6762806541","https://openalex.org/W6766831760","https://openalex.org/W6787972765"],"related_works":["https://openalex.org/W2950520577","https://openalex.org/W1554644772","https://openalex.org/W2003935582","https://openalex.org/W2494130044","https://openalex.org/W3170887803","https://openalex.org/W74409296","https://openalex.org/W1991844655","https://openalex.org/W3209384898","https://openalex.org/W1595834484","https://openalex.org/W2895097035"],"abstract_inverted_index":{"Synchronous":[0],"Stochastic":[1],"Gradient":[2],"Descent":[3],"(SGD)":[4],"with":[5,24],"data":[6,137],"parallelism,":[7,153],"the":[8,49,54,63,70,79,86,96,109,147,177],"most":[9],"popular":[10],"parallel":[11],"training":[12,82,121,197],"strategy":[13,122,198],"for":[14,175],"deep":[15],"learning,":[16],"suffers":[17],"from":[18],"expensive":[19],"gradient":[20],"communications.":[21],"Local":[22],"SGD":[23,81,115],"periodic":[25],"model":[26,50,97,118,134,157,208],"averaging":[27,158],"is":[28,66],"a":[29,132,156,162,172,186,200,206],"promising":[30],"alternative":[31],"to":[32,40,76,107,210],"synchronous":[33,211],"SGD.":[34,212],"The":[35,139],"algorithm":[36,58,159],"allows":[37],"each":[38,129],"worker":[39],"locally":[41],"update":[42],"its":[43],"own":[44],"model,":[45],"and":[46,128],"periodically":[47,143],"averages":[48],"parameters":[51],"across":[52,145],"all":[53,146],"workers.":[55,73],"While":[56],"this":[57,102,151],"enjoys":[59],"less":[60],"frequent":[61],"communications,":[62],"convergence":[64],"rate":[65],"strongly":[67],"affected":[68],"by":[69],"number":[71,87,179],"of":[72,88,111,126,180],"In":[74,101],"order":[75],"scale":[77],"up":[78],"local":[80,114,133,140],"without":[83],"losing":[84],"accuracy,":[85],"workers":[89,181],"should":[90],"be":[91],"sufficiently":[92],"small":[93],"so":[94],"that":[95,160,182,194],"converges":[98],"reasonably":[99],"fast.":[100],"paper,":[103],"we":[104,154],"discuss":[105],"how":[106],"exploit":[108],"degree":[110],"parallelism":[112],"in":[113],"while":[116,204],"maintaining":[117],"accuracy.":[119],"Our":[120,190],"employs":[123],"multiple":[124],"groups":[125],"processes":[127],"group":[130],"trains":[131],"based":[135],"on":[136,150],"parallelism.":[138],"models":[141],"are":[142],"averaged":[144],"groups.":[148],"Based":[149],"hierarchical":[152],"design":[155],"has":[161],"cheaper":[163],"communication":[164],"cost":[165],"than":[166],"allreduce-based":[167],"approach.":[168],"We":[169],"also":[170],"propose":[171],"practical":[173],"metric":[174],"finding":[176],"maximum":[178],"does":[183],"not":[184],"cause":[185],"significant":[187],"accuracy":[188,209],"loss.":[189],"experimental":[191],"results":[192],"demonstrate":[193],"our":[195],"proposed":[196],"provides":[199],"significantly":[201],"improved":[202],"scalability":[203],"achieving":[205],"comparable":[207]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1},{"year":2021,"cited_by_count":2}],"updated_date":"2026-03-28T08:17:26.163206","created_date":"2025-10-10T00:00:00"}