{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,20]],"date-time":"2026-01-20T08:20:57Z","timestamp":1768897257953,"version":"3.49.0"},"reference-count":46,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE\/ACM Trans. Audio Speech Lang. Process."],"published-print":{"date-parts":[[2021]]},"DOI":"10.1109\/taslp.2021.3122349","type":"journal-article","created":{"date-parts":[[2021,10,26]],"date-time":"2021-10-26T20:52:27Z","timestamp":1635281547000},"page":"3565-3576","source":"Crossref","is-referenced-by-count":1,"title":["Asynchronous Decentralized Distributed Training of Acoustic Models"],"prefix":"10.1109","volume":"29","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4865-1307","authenticated-orcid":false,"given":"Xiaodong","family":"Cui","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wei","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Abdullah","family":"Kayi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mingrui","family":"Liu","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ulrich","family":"Finkler","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1343-6837","authenticated-orcid":false,"given":"Brian","family":"Kingsbury","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"George","family":"Saon","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"David","family":"Kung","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref39","first-page":"5330","article-title":"Can decentralized algorithms outperform centralized algorithms? A case study for decentralized parallel stochastic gradient descent","author":"lian","year":"0","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1137\/120880811"},{"key":"ref33","first-page":"1232","article-title":"Large scale distributed deep networks","author":"dean","year":"0","journal-title":"Proc Adv Neural Inf Process Syst (NIPS)"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2013.6638950"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2013.6639348"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2021.05.011"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9054065"},{"key":"ref36","first-page":"3049","article-title":"Asynchronous decentralized parallel stochastic gradient descent","author":"lian","year":"0","journal-title":"Proc 35th Int Conf Mach Learn (ICML)"},{"key":"ref35","article-title":"Nvidia collective communications library (NCCL","year":"2018"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2019-2700"},{"key":"ref10","article-title":"Large batch optimization for deep learning: Training BERT in 76 minutes","author":"you","year":"0","journal-title":"Proc Int Conf Learn Representations (ICLR)"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1145\/364520.364540"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2014-272"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2014-305"},{"key":"ref13","first-page":"235","article-title":"On parallelization of stochastic gradient descent for speech DNNs","author":"seide","year":"0","journal-title":"Proc Int Conf Acoust Speech Signal Process"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2015-354"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2020.2969859"},{"key":"ref16","article-title":"Increasing deep neural network acoustic model size for large vocabulary continuous speech recognition","author":"maas","year":"2014"},{"key":"ref17","article-title":"Deep speech: Scaling up end-to-end speech recognition","author":"hannun","year":"2014"},{"key":"ref18","article-title":"Parallel training of DNNs with netural gradient and parameter averaging","author":"povey","year":"0","journal-title":"Proc Int Conf Learn Representations (ICLR)"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2014.6854672"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053637"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2017-405"},{"key":"ref27","article-title":"ADAM: A method for stochastic optimization","author":"kingma","year":"0","journal-title":"Proc Int Conf Learn Representations (ICLR)"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/TASLP.2017.2756440"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/JETCAS.2019.2912353"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/ASRU.2017.8268933"},{"key":"ref5","first-page":"6670","article-title":"Lessons from building acoustic models with a million hours of speech","author":"parthasarathi","year":"0","journal-title":"Proc Int Conf Acoust Speech Signal Process"},{"key":"ref8","article-title":"Highly scalable deep learning training system with mixed-precision: Training ImageNet in four minutes","author":"jia","year":"0","journal-title":"Proc Workshop Syst ML Open Source Softw Adv Neural Inf Process Syst (NeurIPS)"},{"key":"ref7","article-title":"Accurate, large minibatch SGD: Training ImageNet in 1 hour","author":"goyal","year":"2017"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2012.2205597"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/3225058.3225069"},{"key":"ref1","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"lecun","year":"2015","journal-title":"Nature"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1561\/0100000006"},{"key":"ref20","first-page":"5880","article-title":"Scalable training of deep learning machines by incremental block training with intro-block parallel optimization and blockwise model-update filtering","author":"chen","year":"0","journal-title":"Proc Int Conf Acoust Speech Signal Process"},{"key":"ref45","first-page":"171","article-title":"Model accuracy and runtime tradeoff in distributed deep learning: A systematic study","author":"zhang","year":"0","journal-title":"Proc IEEE Int Conf Data Mining (ICDM)"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2019.8682888"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053993"},{"key":"ref42","first-page":"69","article-title":"The fisher corpus: A resource for the next generation of speech-to-text","author":"cieri","year":"0","journal-title":"Proc ICLRE"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2014-274"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.1992.225858"},{"key":"ref23","first-page":"173","article-title":"Deep speech 2: End-to-end speech recognition in english and mandarin","author":"amodei","year":"0","journal-title":"Proc 35th Int Conf Mach Learn (ICML)"},{"key":"ref44","article-title":"Scale MLPerf-0.6 models on Google TPU-v3 pods","author":"kumar","year":"2019"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9052983"},{"key":"ref43","year":"0"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.24963\/ijcai.2018\/447"}],"container-title":["IEEE\/ACM Transactions on Audio, Speech, and Language Processing"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/6570655\/9289074\/09585552.pdf?arnumber=9585552","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,5,10]],"date-time":"2022-05-10T14:54:02Z","timestamp":1652194442000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9585552\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021]]},"references-count":46,"URL":"https:\/\/doi.org\/10.1109\/taslp.2021.3122349","relation":{},"ISSN":["2329-9290","2329-9304"],"issn-type":[{"value":"2329-9290","type":"print"},{"value":"2329-9304","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021]]}}}