{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T20:31:31Z","timestamp":1767990691147,"version":"3.49.0"},"reference-count":82,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"9","license":[{"start":{"date-parts":[[2022,9,1]],"date-time":"2022-09-01T00:00:00Z","timestamp":1661990400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2022,9,1]],"date-time":"2022-09-01T00:00:00Z","timestamp":1661990400000},"content-version":"am","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2022,9,1]],"date-time":"2022-09-01T00:00:00Z","timestamp":1661990400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2022,9,1]],"date-time":"2022-09-01T00:00:00Z","timestamp":1661990400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CCF-1751356"],"award-info":[{"award-number":["CCF-1751356"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CCF-1956386"],"award-info":[{"award-number":["CCF-1956386"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CNS-0932428"],"award-info":[{"award-number":["CNS-0932428"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CCF-1018927"],"award-info":[{"award-number":["CCF-1018927"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CCF-1423663"],"award-info":[{"award-number":["CCF-1423663"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation","doi-asserted-by":"publisher","award":["CCF-1409204"],"award-info":[{"award-number":["CCF-1409204"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100005144","name":"Qualcomm Inc.","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100005144","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000104","name":"National Aeronautics and Space Administration\u2019s (NASA) Jet Propulsion Laboratory through the President and Director\u2019s Fund","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100000104","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004052","name":"King Abdullah University of Science and Technology","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100004052","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Inform. Theory"],"published-print":{"date-parts":[[2022,9]]},"DOI":"10.1109\/tit.2022.3171173","type":"journal-article","created":{"date-parts":[[2022,4,28]],"date-time":"2022-04-28T20:33:14Z","timestamp":1651177994000},"page":"6078-6097","source":"Crossref","is-referenced-by-count":5,"title":["Differentially Quantized Gradient Methods"],"prefix":"10.1109","volume":"68","author":[{"given":"Chung-Yi","family":"Lin","sequence":"first","affiliation":[{"name":"Kronos Research, Taipei, Taiwan"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2406-7440","authenticated-orcid":false,"given":"Victoria","family":"Kostina","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering, California Institute of Technology, Pasadena, CA, USA"}]},{"given":"Babak","family":"Hassibi","sequence":"additional","affiliation":[{"name":"Kronos Research, Taipei, Taiwan"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/ISIT45174.2021.9518254"},{"key":"ref2","first-page":"2595","article-title":"Parallelized stochastic gradient descent","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Zinkevich"},{"key":"ref3","first-page":"693","article-title":"Hogwild: A lock-free approach to parallelizing stochastic gradient descent","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Recht"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781139042918"},{"key":"ref5","first-page":"1223","article-title":"Large scale distributed deep networks","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Dean"},{"key":"ref6","first-page":"571","article-title":"Project Adam: Building an efficient and scalable deep learning training system","volume-title":"Proc. 11th USENIX Symp. Operating Syst. Des. Implement. (OSDI)","author":"Chilimbi"},{"key":"ref7","first-page":"2674","article-title":"Taming the wild: A unified analysis of hogwild-style algorithms","volume-title":"Advances in Neural Information Processing Systems","author":"De Sa","year":"2015"},{"key":"ref8","first-page":"1","article-title":"Federated learning: Strategies for improving communication efficiency","volume-title":"Proc. NIPS Workshop Private Multi-Party Mach. Learn.","author":"Kone\u010dn\u00fd"},{"key":"ref9","first-page":"3027","article-title":"Optimal algorithms for smooth and strongly convex distributed optimization in networks","volume-title":"Proc. 34th Int. Conf. Mach. Learn.","volume":"70","author":"Scaman"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.21437\/interspeech.2014-274"},{"key":"ref11","first-page":"19","article-title":"Communication efficient distributed machine learning with the parameter server","volume-title":"Proc. Adv. Neural Inf. Process. Syst. (NIPS)","author":"Li"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.21437\/interspeech.2015-354"},{"key":"ref13","first-page":"685","article-title":"Deep learning with elastic averaging SGD","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Zhang"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1214\/aoms\/1177729586"},{"key":"ref15","first-page":"1509","article-title":"TernGrad: Ternary gradients to reduce communication in distributed deep learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Wen"},{"key":"ref16","first-page":"560","article-title":"SignSGD: Compressed optimisation for non-convex problems","volume-title":"Proc. 35th Int. Conf. Mach. Learn.","author":"Bernstein"},{"key":"ref17","first-page":"1709","article-title":"QSGD: Communication-efficient SGD via gradient quantization and encoding","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Alistarh"},{"issue":"114","key":"ref18","first-page":"1","article-title":"Nuqsgd: Improved communication efficiency for data-parallel SGD via nonuniform Quantization","volume":"22","author":"Ramezani-Kebrya","year":"2021","journal-title":"J. Mach. Learn. Res."},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/ISIT44484.2020.9174075"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2022.3161620"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/d17-1045"},{"key":"ref22","first-page":"4447","article-title":"Sparsified SGD with memory","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Stich"},{"key":"ref23","first-page":"1299","article-title":"Gradient sparsification for communication-efficient distributed optimization","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Wangni"},{"key":"ref24","first-page":"9850","article-title":"ATOMO: Communication-efficient learning via atomic sparsification","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Wang"},{"key":"ref25","first-page":"1","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Lin"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/MLHPC.2016.004"},{"key":"ref27","first-page":"5973","article-title":"The convergence of sparsified gradient methods","volume-title":"Advances in Neural Information Processing Systems","author":"Alistarh","year":"2018"},{"key":"ref28","first-page":"5123","article-title":"GradiVeQ: Vector quantization for bandwidth-efficient gradient aggregation in distributed CNN training","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"31","author":"Yu"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/ISIT.2019.8849334"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/TWC.2019.2946245"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/TWC.2019.2961673"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1137\/070704277"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1561\/2200000050"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1137\/16M1080173"},{"key":"ref35","article-title":"Distributed learning with compressed gradient differences","author":"Mishchenko","year":"2019","journal-title":"arXiv:1901.09269"},{"key":"ref36","article-title":"Stochastic distributed learning with gradient quantization and variance reduction","author":"Horv\u00e1th","year":"2019","journal-title":"arXiv:1904.05115"},{"key":"ref37","article-title":"Natural compression for distributed deep learning","author":"Horv\u00e1th","year":"2019","journal-title":"arXiv:1905.10988"},{"key":"ref38","article-title":"Bidirectional compression in heterogeneous settings for distributed or federated learning with partial participation: Tight convergence guarantees","author":"Philippenko","year":"2020","journal-title":"arXiv:2006.14591"},{"key":"ref39","first-page":"680","article-title":"A unified theory of SGD: Variance reduction, sampling, quantization and coordinate descent","volume-title":"Proc. Int. Conf. Artif. Intell. Statist.","author":"Gorbunov"},{"key":"ref40","first-page":"20889","article-title":"Linearly converging error compensated SGD","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Gorbunov"},{"key":"ref41","article-title":"Unified analysis of stochastic gradient methods for composite convex and smooth optimization","author":"Khaled","year":"2020","journal-title":"arXiv:2006.11573"},{"key":"ref42","first-page":"3788","article-title":"MARINA: Faster non-convex distributed learning with compression","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Gorbunov"},{"key":"ref43","first-page":"4617","article-title":"Distributed second order methods with fast rates and compressed communication","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Islamov"},{"key":"ref44","first-page":"3252","article-title":"Error feedback fixes SignSGD and other gradient compression schemes","volume-title":"Proc. 36th Int. Conf. Mach. Learn.","volume":"97","author":"Karimireddy"},{"key":"ref45","article-title":"On biased compression for distributed learning","author":"Beznosikov","year":"2020","journal-title":"arXiv:2002.12410"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2020.3031073"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4613-1643-5"},{"key":"ref48","first-page":"11450","article-title":"Communication-efficient distributed blockwise momentum SGD with error-feedback","volume-title":"Advances in Neural Information Processing Systems","author":"Zheng","year":"2019"},{"key":"ref49","first-page":"5325","article-title":"Error compensated quantized SGD and its applications to large-scale distributed optimization","volume-title":"Proc. 35th Int. Conf. Mach. Learn.","volume":"80","author":"Wu"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2020.3033286"},{"key":"ref51","first-page":"1","article-title":"Error compensated distributed SGD can be accelerated","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"34","author":"Qian"},{"key":"ref52","first-page":"1","article-title":"EF21: A new, simpler, theoretically better, and practically faster error feedback","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"34","author":"Richt\u00e1rik"},{"key":"ref53","first-page":"1","article-title":"A better alternative to error feedback for communication-efficient distributed learning","volume-title":"Proc. 8th Int. Conf. Learn. Represent.","author":"Horv\u00e1th"},{"key":"ref54","first-page":"40","article-title":"Distributed learning with sublinear communication","volume-title":"Proc. 36th Int. Conf. Mach. Learn.","volume":"97","author":"Acharya"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2021.3058663"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1561\/2200000016"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/CDC.2012.6426691"},{"key":"ref58","first-page":"3329","article-title":"Distributed mean estimation with limited communication","volume-title":"Proc. 34th Int. Conf. Mach. Learn.","volume":"70","author":"Suresh"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1109\/TAC.2009.2031203"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2019.2932876"},{"key":"ref61","first-page":"2328","article-title":"Information-theoretic lower bounds for distributed statistical estimation with communication constraints","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Zhang"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.1016\/0885-064X(87)90013-6"},{"key":"ref63","first-page":"1756","article-title":"Communication complexity of distributed convex learning and optimization","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Arjevani"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.5555\/2685048.2685095"},{"key":"ref65","article-title":"Distributed learning with compressed gradients","author":"Khirirat","year":"2018","journal-title":"arXiv:1806.06573"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4419-8853-9"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1017\/9781316227268.003"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1112\/S0025579300004083"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1137\/110830629"},{"issue":"2","key":"ref70","first-page":"372","article-title":"A method of solving a convex programming problem with convergence rate O(1\/k2)","volume":"27","author":"Nesterov","year":"1982","journal-title":"Sov. Math. Doklady"},{"key":"ref71","article-title":"Linear coupling: An ultimate unification of gradient and mirror descent","volume-title":"Proc. 8th Innov. Theor. Comput. Sci. Conf.","author":"Allen-Zhu"},{"issue":"153","key":"ref72","first-page":"1","article-title":"A differential equation for modeling Nesterov\u2019s accelerated gradient method: Theory and insights","volume":"17","author":"Su","year":"2016","journal-title":"J. Mach. Learn. Res."},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781139045520"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.21105\/joss.01244"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1007\/s11590-016-1087-4"},{"issue":"126","key":"ref76","first-page":"1","article-title":"On lower and upper bounds in smooth and strongly convex optimization","volume":"17","author":"Arjevani","year":"2016","journal-title":"J. Mach. Learn. Res."},{"key":"ref77","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9780511804441"},{"key":"ref78","volume-title":"Learning From Data","author":"Abu-Mostafa","year":"2012"},{"key":"ref79","article-title":"Fast convergence of stochastic gradient descent under a strong growth condition","author":"Schmidt","year":"2013","journal-title":"arXiv:1308.6370"},{"key":"ref80","first-page":"1017","article-title":"Stochastic gradient descent, weighted sampling, and the randomized Kaczmarz algorithm","volume-title":"Advances in Neural Information Processing Systems","volume":"27","author":"Needell","year":"2014"},{"key":"ref81","first-page":"3325","article-title":"The power of interpolation: Understanding the effectiveness of SGD in modern over-parametrized learning","volume-title":"Proc. Int. Conf. Mach. Learn.","volume":"80","author":"Ma"},{"key":"ref82","doi-asserted-by":"publisher","DOI":"10.1002\/(SICI)1099-1239(1998100)8:12<1043::AID-RNC364>3.0.CO;2-H"}],"container-title":["IEEE Transactions on Information Theory"],"original-title":[],"link":[{"URL":"https:\/\/ieeexplore.ieee.org\/ielam\/18\/9861208\/9764884-aam.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/18\/9861208\/09764884.pdf?arnumber=9764884","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,1,22]],"date-time":"2024-01-22T22:38:18Z","timestamp":1705963098000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9764884\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,9]]},"references-count":82,"journal-issue":{"issue":"9"},"URL":"https:\/\/doi.org\/10.1109\/tit.2022.3171173","relation":{},"ISSN":["0018-9448","1557-9654"],"issn-type":[{"value":"0018-9448","type":"print"},{"value":"1557-9654","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,9]]}}}