{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,13]],"date-time":"2026-10-13T17:28:50Z","timestamp":1791912530584,"version":"4.3.4"},"reference-count":48,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"1","license":[{"start":{"date-parts":[[2020,5,1]],"date-time":"2020-05-01T00:00:00Z","timestamp":1588291200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2020,5,1]],"date-time":"2020-05-01T00:00:00Z","timestamp":1588291200000},"content-version":"am","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2020,5,1]],"date-time":"2020-05-01T00:00:00Z","timestamp":1588291200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2020,5,1]],"date-time":"2020-05-01T00:00:00Z","timestamp":1588291200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100008982","name":"NSF","doi-asserted-by":"publisher","award":["1514531"],"award-info":[{"award-number":["1514531"]}],"id":[{"id":"10.13039\/501100008982","id-type":"DOI","asserted-by":"publisher"}]},{"name":"UC-NL","award":["LFR-18-548554"],"award-info":[{"award-number":["LFR-18-548554"]}]},{"DOI":"10.13039\/100006754","name":"Army Research Laboratory under Cooperative Agreement","doi-asserted-by":"publisher","award":["W911NF-17-2-0196"],"award-info":[{"award-number":["W911NF-17-2-0196"]}],"id":[{"id":"10.13039\/100006754","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE J. Sel. Areas Inf. Theory"],"published-print":{"date-parts":[[2020,5]]},"DOI":"10.1109\/jsait.2020.2985917","type":"journal-article","created":{"date-parts":[[2020,4,6]],"date-time":"2020-04-06T21:41:07Z","timestamp":1586209267000},"page":"217-226","source":"Crossref","is-referenced-by-count":159,"title":["Qsparse-Local-SGD: Distributed SGD With Quantization, Sparsification, and Local Computations"],"prefix":"10.1109","volume":"1","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-3130-4253","authenticated-orcid":false,"given":"Debraj","family":"Basu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3544-8414","authenticated-orcid":false,"given":"Deepesh","family":"Data","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Can","family":"Karakus","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Suhas N.","family":"Diggavi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1145\/1273496.1273598"},{"key":"ref38","first-page":"1","article-title":"Adam: A method for stochastic optimization","author":"kingma","year":"2015","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref33","first-page":"3478","article-title":"Decentralized stochastic optimization and Gossip algorithms with compressed communication","author":"koloskova","year":"2019","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref32","first-page":"3252","article-title":"Error feedback fixes SignSGD and other gradient compression schemes","author":"karimireddy","year":"2019","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1137\/16M1057000"},{"key":"ref30","author":"zhang","year":"2016","journal-title":"Parallel SGD When does averaging help?"},{"key":"ref37","first-page":"26","article-title":"Lecture 6.5-RMSprop: Divide the gradient by a running average of its recent magnitude","volume":"4","author":"tieleman","year":"2012","journal-title":"COURSERA Neural Netw Mach Learn"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/ICNN.1993.298623"},{"key":"ref35","first-page":"2328","article-title":"Information-theoretic lower bounds for distributed statistical estimation with communication constraints","author":"zhang","year":"2013","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"ref10","author":"konecn\u00fd","year":"2017","journal-title":"Stochastic distributed and federated optimization for machine learning"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1137\/070704277"},{"key":"ref11","first-page":"1273","article-title":"Communication-efficient learning of deep networks from decentralized data","author":"mcmahan","year":"2017","journal-title":"Proc Artif Intell Statist (AISTATS) Conf"},{"key":"ref12","first-page":"265","article-title":"TensorFlow: A system for large-scale machine learning","author":"abadi","year":"2016","journal-title":"Proc of the 2nd USENIX Symp on Operating Systems Design and Implementation (OSDI)"},{"key":"ref13","first-page":"1707","article-title":"QSGD: Communication-efficient SGD via gradient quantization and encoding","author":"alistarh","year":"2017","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref14","first-page":"5321","article-title":"Error compensated quantized SGD and its applications to large-scale distributed optimization","author":"wu","year":"2018","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref15","first-page":"5811","article-title":"Training quantized nets: A deeper understanding","author":"li","year":"2017","journal-title":"Proc Neural Inf Process Syst (NIPS)"},{"key":"ref16","first-page":"1488","article-title":"Scalable distributed DNN training using commodity GPU cloud computing","author":"strom","year":"2015","journal-title":"Proc Conf Int Speech Commun Assoc (INTERSPEECH)"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D17-1045"},{"key":"ref18","first-page":"4452","article-title":"Sparsified SGD with memory","author":"stich","year":"2018","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref19","first-page":"1","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training","author":"lin","year":"2018","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref28","first-page":"1306","article-title":"Gradient sparsification for communication-efficient distributed optimization","author":"wangni","year":"2018","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref4","first-page":"451","article-title":"Non-asymptotic analysis of stochastic approximation algorithms for machine learning","author":"bach","year":"2011","journal-title":"Proc Neural Inf Process Syst (NIPS)"},{"key":"ref27","first-page":"1058","article-title":"1-bit stochastic gradient descent and its application to data-parallel distributed training of speech DNNs","author":"seide","year":"2014","journal-title":"Proc INTERSPEECH"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-7908-2604-3_16"},{"key":"ref6","first-page":"1508","article-title":"TernGrad: Ternary gradients to reduce communication in distributed deep learning","author":"wen","year":"2017","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref29","first-page":"9872","article-title":"ATOMO: Communication-efficient learning via atomic sparsification","author":"wang","year":"2018","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref5","first-page":"5977","article-title":"The convergence of sparsified gradient methods","author":"alistarh","year":"2018","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref8","first-page":"3329","article-title":"Distributed mean estimation with limited communication","author":"suresh","year":"2017","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref7","first-page":"559","article-title":"SignSGD: Compressed optimisation for non-convex problems","author":"bernstein","year":"2018","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref2","doi-asserted-by":"crossref","first-page":"400","DOI":"10.1214\/aoms\/1177729586","article-title":"A stochastic approximation method","volume":"22","author":"herbert","year":"1951","journal-title":"Ann Math Stat"},{"key":"ref9","author":"devlin","year":"2018","journal-title":"BERT Pre-training of deep bidirectional transformers for language understanding"},{"key":"ref1","first-page":"14668","article-title":"QSparse-local-SGD: Distributed SGD with quantization, sparsification and local computations","author":"basu","year":"2019","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v33i01.33015693"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/TSIPN.2017.2695121"},{"key":"ref48","first-page":"325","article-title":"Top-k multiclass SVM","author":"lapin","year":"2015","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref22","first-page":"3321","article-title":"Communication-efficient algorithms for statistical optimization","volume":"14","author":"zhang","year":"2013","journal-title":"J Mach Learn Res"},{"key":"ref47","author":"sergeev","year":"2018","journal-title":"Horovod fast and easy distributed deep learning in tensorflow"},{"key":"ref21","article-title":"Iterative parameter mixing for distributed large-margin training of structured predictors for natural language processing","author":"coppola","year":"2015"},{"key":"ref42","first-page":"2489","article-title":"Beyond the regret minimization barrier: Optimal algorithms for stochastic strongly-convex optimization","volume":"15","author":"hazan","year":"2014","journal-title":"J Mach Learn Res"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2016.7472805"},{"key":"ref41","first-page":"693","article-title":"HogWild: A lock-free approach to parallelizing stochastic gradient descent","author":"recht","year":"2011","journal-title":"Proc Neural Inf Process Syst"},{"key":"ref23","first-page":"1","article-title":"Local SGD converges fast and communicates little","author":"stich","year":"2019","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref44","first-page":"3747","article-title":"SGD and Hogwild! Convergence without the bounded gradients assumption","author":"nguyen","year":"2018","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/TCT.1973.1083627"},{"key":"ref43","first-page":"1","article-title":"Making gradient descent optimal for strongly convex stochastic optimization","author":"rakhlin","year":"2012","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref25","author":"wang","year":"2018","journal-title":"Cooperative SGD A unified framework for the design and analysis of communication-efficient SGD algorithms"}],"container-title":["IEEE Journal on Selected Areas in Information Theory"],"original-title":[],"link":[{"URL":"https:\/\/ieeexplore.ieee.org\/ielam\/8700143\/8768428\/9057579-aam.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/8700143\/8768428\/09057579.pdf?arnumber=9057579","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,4,27]],"date-time":"2022-04-27T09:17:32Z","timestamp":1651051052000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9057579\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2020,5]]},"references-count":48,"journal-issue":{"issue":"1"},"URL":"https:\/\/doi.org\/10.1109\/jsait.2020.2985917","relation":{},"ISSN":["2641-8770"],"issn-type":[{"value":"2641-8770","type":"electronic"}],"subject":[],"published":{"date-parts":[[2020,5]]}}}