{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,30]],"date-time":"2026-03-30T21:01:04Z","timestamp":1774904464591,"version":"3.50.1"},"reference-count":77,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62472050"],"award-info":[{"award-number":["62472050"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62473146"],"award-info":[{"award-number":["62473146"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004735","name":"Natural Science Foundation of Hunan Province","doi-asserted-by":"publisher","award":["2024JJ3017"],"award-info":[{"award-number":["2024JJ3017"]}],"id":[{"id":"10.13039\/501100004735","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004735","name":"Natural Science Foundation of Hunan Province","doi-asserted-by":"publisher","award":["2025JJ20070"],"award-info":[{"award-number":["2025JJ20070"]}],"id":[{"id":"10.13039\/501100004735","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100016716","name":"Hong Kong RGC TRS","doi-asserted-by":"publisher","award":["T41-603\/20R"],"award-info":[{"award-number":["T41-603\/20R"]}],"id":[{"id":"10.13039\/100016716","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Netw."],"published-print":{"date-parts":[[2026]]},"DOI":"10.1109\/ton.2026.3673659","type":"journal-article","created":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T19:52:30Z","timestamp":1773431550000},"page":"4219-4232","source":"Crossref","is-referenced-by-count":0,"title":["Towards Optimal Communication Scheduling With Automatic Configuration for Distributed DNN Training"],"prefix":"10.1109","volume":"34","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8216-9683","authenticated-orcid":false,"given":"Jinbin","family":"Hu","sequence":"first","affiliation":[{"name":"Changsha University of Science and Technology., Changsha, China"}]},{"given":"Xinming","family":"Xu","sequence":"additional","affiliation":[{"name":"Changsha University of Science and Technology., Changsha, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9883-2400","authenticated-orcid":false,"given":"Hao","family":"Wang","sequence":"additional","affiliation":[{"name":"The Hong Kong University of Science and Technology, Hong Kong, China"}]},{"given":"Yiqing","family":"Ma","sequence":"additional","affiliation":[{"name":"The Hong Kong University of Science and Technology, Hong Kong, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5910-0661","authenticated-orcid":false,"given":"Xiaolin","family":"Chen","sequence":"additional","affiliation":[{"name":"Qiongtai Normal University., Haikou, China"}]},{"given":"Yiming","family":"Zhang","sequence":"additional","affiliation":[{"name":"Shanghai Jiao Tong University., Shanghai, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5473-8738","authenticated-orcid":false,"given":"Jin","family":"Wang","sequence":"additional","affiliation":[{"name":"Hunan University of Science and Technology., Xiangtan, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2587-6028","authenticated-orcid":false,"given":"Kai","family":"Chen","sequence":"additional","affiliation":[{"name":"The Hong Kong University of Science and Technology, Hong Kong, China"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM48880.2022.9796752"},{"key":"ref2","first-page":"1","article-title":"Accelerating skewed workloads with performance multipliers in the TurboDB distributed database","volume-title":"Proc. USENIX NSDI","author":"Lam"},{"key":"ref3","first-page":"1","article-title":"Enabling tensor language model to assist in generating high-performance tensor programs for deep learning","volume-title":"Proc. USENIX OSDI","author":"Zhai"},{"key":"ref4","first-page":"1","article-title":"Parcae: Proactive,liveput-optimized DNN training on preemptible instances","volume-title":"Proc. USENIX NSDI","author":"Duan"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3718958.3750503"},{"key":"ref6","first-page":"1","article-title":"TopoOpt: Co-optimizing network topology and parallelization strategy for distributed training jobs","volume-title":"Proc. USENIX NSDI","author":"Weiyang"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3627703.3629554"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3552326.3567499"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/3651890.3672249"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1145\/3718958.3750514"},{"key":"ref11","first-page":"1","article-title":"Pollux: Co-adaptive cluster scheduling for goodput-optimized deep learning","volume-title":"Proc. USENIX OSDI","author":"Qiao"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3663408.3663409"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/3694715.3695969"},{"key":"ref14","first-page":"579","article-title":"FwdLLM: Efficient federated finetuning of large language models with perturbed inferences","volume-title":"Proc. USENIX ATC","author":"Xu"},{"key":"ref15","first-page":"1","article-title":"Metis: Fast automatic distributed training on heterogeneous GPUs","volume-title":"Proc. USENIX ATC","author":"Um"},{"key":"ref16","article-title":"MXNet: A flexible and efficient machine learning library for heterogeneous distributed systems","author":"Chen","year":"2015","journal-title":"arXiv:1512.01274"},{"key":"ref17","first-page":"1","article-title":"Bamboo: Making preemptible instances resilient for affordable training of large DNNs","volume-title":"Proc. USENIX NSDI","author":"Thorpe"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/TNET.2024.3414853"},{"key":"ref19","first-page":"1","article-title":"TicTac: Accelerating distributed deep learning with communication scheduling","volume-title":"Proc. MLSys","author":"Hashemi"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1145\/3190508.3190517"},{"key":"ref21","article-title":"Biased local SGD for efficient deep learning on heterogeneous systems","author":"Lim","year":"2025","journal-title":"arXiv:2508.08540"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/TNET.2023.3339524"},{"key":"ref23","first-page":"1","article-title":"Priority-based parameter propagation for distributed DNN training","volume-title":"Proc. MLSys","author":"Jayarajan"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3627703.3650083"},{"key":"ref25","first-page":"1","article-title":"A unified architecture for accelerating distributed DNN training in heterogeneous GPU\/CPU clusters","volume-title":"Proc. USENIX OSDI","author":"Jiang"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.32604\/jai.2024.054314"},{"key":"ref27","first-page":"1","article-title":"CASSINI: Network-aware job scheduling in machine learning clusters","volume-title":"Proc. USENIX NSDI","author":"Rajasekaran"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/3065386"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/3575693.3575702"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1145\/3694715.3695960"},{"key":"ref31","article-title":"PyTorch: An imperative style, high-performance deep learning library","author":"Paszke","year":"2019","journal-title":"arXiv:1912.01703"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359642"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM52122.2024.10621348"},{"key":"ref34","article-title":"A tutorial on Bayesian optimization of expensive cost functions, with application to active user modeling and hierarchical reinforcement learning","author":"Brochu","year":"2010","journal-title":"arXiv:1012.2599"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3694715.3695961"},{"key":"ref36","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014","journal-title":"arXiv:1409.1556"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1145\/3694715.3695978"},{"key":"ref38","first-page":"11525","article-title":"Dash: Semi-supervised learning with dynamic thresholding","volume-title":"Proc. ICML","author":"Xu"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1145\/3552326.3587445"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/TNET.2024.3366336"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.5555\/3026877.3026899"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1109\/TNET.2023.3244794"},{"key":"ref43","first-page":"1","article-title":"Poseidon: An efficient communication architecture for distributed deep learning on GPU clusters","volume-title":"Proc. USENIX ATC","author":"Zhang"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1145\/3600006.3613142"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1016\/j.inffus.2023.101869"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM53939.2023.10228964"},{"key":"ref49","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training","author":"Lin","year":"2017","journal-title":"arXiv:1712.01887"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1016\/j.jpdc.2008.09.002"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM53939.2023.10229076"},{"key":"ref52","first-page":"1","article-title":"Themis: Fair and efficient GPU cluster scheduling","volume-title":"Proc. USENIX NSDI","author":"Mahajan"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1007\/s11023-020-09548-1"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/26.1.73"},{"key":"ref55","article-title":"Accurate, large minibatch SGD: Training ImageNet in 1 hour","author":"Goyal","year":"2017","journal-title":"arXiv:1706.02677"},{"key":"ref56","article-title":"DenseNet: Implementing efficient ConvNet descriptor pyramids","author":"Iandola","year":"2014","journal-title":"arXiv:1404.1869"},{"key":"ref57","volume-title":"NCCL 2.0","author":"Jeaugey","year":"2017"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1145\/3651890.3672239"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1145\/2640087.2644155"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1145\/3641289"},{"key":"ref61","article-title":"Local SGD converges fast and communicates little","author":"Stich","year":"2018","journal-title":"arXiv:1805.09767"},{"key":"ref62","first-page":"1","article-title":"QSGD: Communication-efficient SGD via gradient quantization and encoding","volume-title":"Proc. NIPS","author":"Alistarh"},{"key":"ref63","first-page":"1","article-title":"More effective distributed ml via a stale synchronous parallel parameter server","volume-title":"Proc. NIPS","author":"Ho"},{"key":"ref64","first-page":"1","article-title":"Shockwave: Fair and efficient cluster scheduling for dynamic adaptation in machine learning","volume-title":"Proc. USENIX NSDI","author":"Zheng"},{"key":"ref65","volume-title":"ZeroMQ: Messaging for Many Applications","author":"Hintjens","year":"2013"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1145\/2934872.2934908"},{"key":"ref67","first-page":"741","article-title":"ATP: In-network aggregation for multi-tenant learning","volume-title":"Proc. USENIX NSDI","author":"Lao"},{"key":"ref68","first-page":"469","article-title":"CherryPick: Adaptively unearthing the best cloud configurations for big data analytics","volume-title":"Proc. USENIX NSDI","author":"Alipourfard"},{"key":"ref69","first-page":"1","article-title":"Practical Bayesian optimization of machine learning algorithms","volume-title":"Proc. NeurIPS","author":"Snoek"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.14778\/1687627.1687767"},{"key":"ref71","first-page":"1","article-title":"PCC: Re-architecting congestion control for consistent high performance","volume-title":"Proc. USENIX NSDI","author":"Dong"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1145\/3230543.3230551"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2020.106622"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.1145\/3544216.3544224"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1145\/3470496.3533727"},{"key":"ref76","article-title":"Model parallelism on distributed infrastructure: A literature review from theory to LLM case-studies","author":"Brakel","year":"2024","journal-title":"arXiv:2403.03699"},{"key":"ref77","first-page":"1","article-title":"Towards domain-specific network transport for distributed DNN training","volume-title":"Proc. USENIX NSDI","author":"Wang"}],"container-title":["IEEE Transactions on Networking"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10723154\/11317935\/11434550.pdf?arnumber=11434550","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,30]],"date-time":"2026-03-30T20:05:54Z","timestamp":1774901154000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11434550\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":77,"URL":"https:\/\/doi.org\/10.1109\/ton.2026.3673659","relation":{},"ISSN":["2998-4157"],"issn-type":[{"value":"2998-4157","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026]]}}}