{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T15:53:09Z","timestamp":1783439589355,"version":"3.54.6"},"reference-count":122,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"8","license":[{"start":{"date-parts":[[2019,8,1]],"date-time":"2019-08-01T00:00:00Z","timestamp":1564617600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/OAPA.html"}],"funder":[{"DOI":"10.13039\/501100012166","name":"National Basic Research Program of China","doi-asserted-by":"publisher","award":["2017YFB1001703"],"award-info":[{"award-number":["2017YFB1001703"]}],"id":[{"id":"10.13039\/501100012166","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["U1711265"],"award-info":[{"award-number":["U1711265"]}],"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":["61802449"],"award-info":[{"award-number":["61802449"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Program for Guangdong Introducing Innovative and Entrepreneurial Teams","award":["2017ZT07X355"],"award-info":[{"award-number":["2017ZT07X355"]}]},{"name":"Guangdong Natural Science Funds","award":["2018A030313032"],"award-info":[{"award-number":["2018A030313032"]}]},{"DOI":"10.13039\/501100012226","name":"Fundamental Research Funds for the Central Universities","doi-asserted-by":"crossref","award":["17lgjc40"],"award-info":[{"award-number":["17lgjc40"]}],"id":[{"id":"10.13039\/501100012226","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100000183","name":"Army Research Office","doi-asserted-by":"publisher","award":["W911NF-16-1-0448"],"award-info":[{"award-number":["W911NF-16-1-0448"]}],"id":[{"id":"10.13039\/100000183","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000774","name":"Defense Threat Reduction Agency","doi-asserted-by":"publisher","award":["HDTRA1-13-1-0029"],"award-info":[{"award-number":["HDTRA1-13-1-0029"]}],"id":[{"id":"10.13039\/100000774","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Proc. IEEE"],"published-print":{"date-parts":[[2019,8]]},"DOI":"10.1109\/jproc.2019.2918951","type":"journal-article","created":{"date-parts":[[2019,6,12]],"date-time":"2019-06-12T19:52:43Z","timestamp":1560369163000},"page":"1738-1762","source":"Crossref","is-referenced-by-count":2043,"title":["Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing"],"prefix":"10.1109","volume":"107","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0987-9344","authenticated-orcid":false,"given":"Zhi","family":"Zhou","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9943-6020","authenticated-orcid":false,"given":"Xu","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9928-5813","authenticated-orcid":false,"given":"En","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4800-8768","authenticated-orcid":false,"given":"Liekang","family":"Zeng","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0118-7236","authenticated-orcid":false,"given":"Ke","family":"Luo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3840-1753","authenticated-orcid":false,"given":"Junshan","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref39","article-title":"A Berkeley view of systems challenges for AI","author":"stoica","year":"2017","journal-title":"arXiv 1712 05855"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/SEC.2018.00016"},{"key":"ref33","year":"0","journal-title":"3 AI Trends for Enterprise Computing"},{"key":"ref32","doi-asserted-by":"crossref","first-page":"529","DOI":"10.1038\/nature14236","article-title":"Human-level control through deep reinforcement learning","volume":"518","author":"mnih","year":"2015","journal-title":"Nature"},{"key":"ref31","first-page":"2672","article-title":"Generative adversarial nets","author":"goodfellow","year":"2014","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.8.1735"},{"key":"ref37","first-page":"377","article-title":"Live video analytics at scale with approximation and delay-tolerance","author":"zhang","year":"2017","journal-title":"Proc USENIX NSDI"},{"key":"ref36","year":"0","journal-title":"Microsoft Interactive Cloud Gaming"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/MPRV.2009.82"},{"key":"ref34","year":"0","journal-title":"Democratizing AI"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2015.7298594"},{"key":"ref27","doi-asserted-by":"crossref","DOI":"10.4324\/9780203763247","author":"chauvin","year":"2013","journal-title":"Backpropagation Theory Architectures and Applications"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/5.58337"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/TETC.2016.2593643"},{"key":"ref21","article-title":"MobileNets: Efficient convolutional neural networks for mobile vision applications","author":"howard","year":"2017","journal-title":"arXiv 1704 04861"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-46448-0_2"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.91"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1038\/323533a0"},{"key":"ref101","article-title":"Picking interesting frames in streaming video","author":"canel","year":"0"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-7908-2604-3_16"},{"key":"ref100","doi-asserted-by":"publisher","DOI":"10.1145\/3225058.3225103"},{"key":"ref50","first-page":"629","article-title":"Gaia: Geo-distributed machine learning approaching LAN speeds","author":"hsieh","year":"2017","journal-title":"Proc NSDI"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/JSAC.2019.2904348"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1145\/3219819.3220106"},{"key":"ref58","article-title":"A privacy-preserving deep learning approach for face recognition with edge computing","author":"mao","year":"2018","journal-title":"Proc USENIX"},{"key":"ref57","article-title":"Machine learning at the wireless edge: Distributed stochastic gradient descent over-the-air","author":"amiri","year":"2019","journal-title":"arXiv 1901 00844"},{"key":"ref56","first-page":"7663","article-title":"Communication compression for decentralized training","author":"tang","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref55","first-page":"4452","article-title":"Sparsified sgd with memory","author":"stich","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref54","article-title":"eSGD: Communication efficient distributed deep learning on the edge","author":"tao","year":"2018","journal-title":"Proceedings of USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18)"},{"key":"ref53","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training","author":"lin","year":"2017","journal-title":"arXiv 1712 01887"},{"key":"ref52","article-title":"Client selection for federated learning with heterogeneous resources in mobile edge","author":"nishio","year":"2018","journal-title":"arXiv 1804 08333"},{"key":"ref40","year":"2018","journal-title":"5 Trends Emerge in the Gartner Hype Cycle for Emerging Technologies"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/2749246.2749276"},{"key":"ref3","year":"0","journal-title":"Cisco global cloud index Forecast and methodology 2016–2021"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2016.2579198"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/2785956.2787505"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/COMST.2017.2745201"},{"key":"ref49","article-title":"On-device federated learning via blockchain and its latency analysis","author":"kim","year":"2018","journal-title":"arXiv 1808 03949"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/MWC.2017.1600321"},{"key":"ref9","article-title":"In-edge Ai: Intelligentizing mobile edge computing, caching and communication by federated learning","author":"wang","year":"2018","journal-title":"arXiv 1809 07857"},{"key":"ref46","doi-asserted-by":"crossref","first-page":"1310","DOI":"10.1145\/2810103.2813687","article-title":"Privacy-preserving deep learning","author":"shokri","year":"2015","journal-title":"Proc 22nd ACM SIGSAC Conf Comput Commun Secur"},{"key":"ref45","article-title":"Communication-efficient learning of deep networks from decentralized data","author":"mcmahan","year":"2016","journal-title":"arXiv 1602 05629"},{"key":"ref48","article-title":"Peer-to-peer federated learning on graphs","author":"lalitha","year":"2019","journal-title":"arXiv 1901 11173"},{"key":"ref47","article-title":"Federated learning: Strategies for improving communication efficiency","author":"kone?n\u00fd","year":"2016","journal-title":"arXiv 1610 05492"},{"key":"ref42","year":"0","journal-title":"Accelerating AI on the Intelligent Edge"},{"key":"ref41","year":"2017","journal-title":"IEC White Paper Edge Intelligence"},{"key":"ref44","article-title":"Data-driven task allocation for multi-task transfer learning on the edge","author":"chen","year":"2019","journal-title":"Proc IEEE 39th Int Conf Distrib Comput Syst (ICDCS)"},{"key":"ref43","year":"0","journal-title":"Edge Intelligence for Industrial Internet of Things"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.643"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001177"},{"key":"ref71","article-title":"Deep compression: Compressing deep neural networks with pruning, trained quantization and Huffman coding","author":"han","year":"2015","journal-title":"arXiv 1510 00149 [cs]"},{"key":"ref70","first-page":"1135","article-title":"Learning both weights and connections for efficient neural network","author":"han","year":"2015","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref76","doi-asserted-by":"publisher","DOI":"10.1145\/3243176.3243180"},{"key":"ref77","doi-asserted-by":"publisher","DOI":"10.1109\/IPSN.2016.7460664"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001165"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1145\/3210240.3210337"},{"key":"ref78","doi-asserted-by":"crossref","DOI":"10.1109\/MNET.001.1800506","article-title":"Boomerang: On-demand cooperative deep neural network inference for edge intelligence on industrial Internet of Things","author":"zeng","year":"2019","journal-title":"IEEE Netw"},{"key":"ref79","doi-asserted-by":"publisher","DOI":"10.1145\/3093336.3037698"},{"key":"ref60","article-title":"A hybrid deep learning architecture for privacy-preserving mobile analytics","author":"osia","year":"2017","journal-title":"arXiv 1703 02952"},{"key":"ref62","doi-asserted-by":"crossref","first-page":"42","DOI":"10.1109\/MIC.2018.112102519","article-title":"Are existing knowledge transfer techniques effective for deep learning with edge devices?","author":"sharma","year":"2018","journal-title":"Proc IEEE Int Conf Edge Comput (EDGE)"},{"key":"ref61","article-title":"PipeDream: Fast and efficient pipeline parallel DNN training","author":"harlap","year":"2018","journal-title":"arXiv 1806 03377"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2006.874516"},{"key":"ref64","article-title":"Gossip training for deep learning","author":"blot","year":"2016","journal-title":"arXiv 1611 09726"},{"key":"ref65","article-title":"How to scale distributed deep learning?","author":"jin","year":"2016","journal-title":"arXiv 1611 04581"},{"key":"ref66","article-title":"GossipGraD: Scalable deep learning using gossip communication based asynchronous gradient descent","author":"daily","year":"2018","journal-title":"arXiv 1803 05880"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1145\/3278721.3278778"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO.2018.00023"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2019.00048"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1561\/2000000039"},{"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":"ref95","doi-asserted-by":"publisher","DOI":"10.1145\/3241539.3241557"},{"key":"ref109","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-94295-7_2"},{"key":"ref94","first-page":"17","article-title":"Precog: Prefetching for image recognition applications at the edge","author":"drolia","year":"2017","journal-title":"ACM\/IEEE Symposium on Edge Computing (SEC)"},{"key":"ref108","doi-asserted-by":"publisher","DOI":"10.1145\/3081333.3081359"},{"key":"ref93","doi-asserted-by":"publisher","DOI":"10.1109\/ICDCS.2017.94"},{"key":"ref107","doi-asserted-by":"publisher","DOI":"10.1145\/3241539.3241559"},{"key":"ref92","doi-asserted-by":"publisher","DOI":"10.1145\/2809695.2809711"},{"key":"ref106","doi-asserted-by":"publisher","DOI":"10.1145\/3240765.3240796"},{"key":"ref91","doi-asserted-by":"publisher","DOI":"10.1007\/s10115-017-1029-1"},{"key":"ref105","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2018.2880196"},{"key":"ref90","doi-asserted-by":"publisher","DOI":"10.1109\/ICCD.2017.49"},{"key":"ref104","doi-asserted-by":"publisher","DOI":"10.1145\/3230543.3230574"},{"key":"ref103","doi-asserted-by":"publisher","DOI":"10.1145\/3211332.3211336"},{"key":"ref102","doi-asserted-by":"publisher","DOI":"10.1109\/CODESISSS.2015.7331375"},{"key":"ref111","article-title":"Accelerating deep learning workloads through efficient multi-model execution","author":"narayanan","year":"2018","journal-title":"Proc NIPS Workshop Mach Learn Syst"},{"key":"ref112","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM.2018.8485905"},{"key":"ref110","first-page":"29","article-title":"Mainstream: Dynamic stem-sharing for multi-tenant video processing","author":"jiang","year":"2018","journal-title":"Proc USENIX ATC"},{"key":"ref98","doi-asserted-by":"publisher","DOI":"10.1109\/SEC.2018.00019"},{"key":"ref99","article-title":"ReXCam: Resource-efficient, cross-camera video analytics at enterprise scale","author":"jain","year":"2018","journal-title":"arXiv 1811 01268"},{"key":"ref96","article-title":"Shadow puppets: Cloud-level accurate AI inference at the speed and economy of edge","author":"venugopal","year":"2018","journal-title":"Proceedings of USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18)"},{"key":"ref97","doi-asserted-by":"publisher","DOI":"10.14778\/3137628.3137664"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1145\/3229556.3229562"},{"key":"ref11","year":"2018","journal-title":"5 Trends Emerge in the Gartner Hype Cycle for Emerging Technologies"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2017.3641638"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/2594368.2594383"},{"key":"ref14","first-page":"1756","article-title":"EdgeOS_h: A home operating system for Internet of everything","author":"jie","year":"2017","journal-title":"Proc IEEE ICDCS"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/TII.2018.2842821"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1016\/S0169-7439(97)00061-0"},{"key":"ref82","doi-asserted-by":"publisher","DOI":"10.1145\/3267809.3267828"},{"key":"ref118","first-page":"3368","article-title":"Gradient coding: Avoiding stragglers in distributed learning","author":"tandon","year":"2017","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref17","first-page":"2493","article-title":"Natural language processing (almost) from scratch","volume":"12","author":"collobert","year":"2011","journal-title":"J Mach Learn Res"},{"key":"ref81","doi-asserted-by":"publisher","DOI":"10.1109\/PADSW.2018.8645013"},{"key":"ref117","article-title":"Neural architecture search with reinforcement learning","author":"zoph","year":"2016","journal-title":"arXiv 1611 01578"},{"key":"ref18","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","author":"krizhevsky","year":"2012","journal-title":"Proc NIPS"},{"key":"ref84","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD.2017.8203852"},{"key":"ref19","article-title":"Very deep convolutional networks for large-scale image recognition","author":"simonyan","year":"2014","journal-title":"arXiv 1409 1556"},{"key":"ref83","doi-asserted-by":"publisher","DOI":"10.1109\/AVSS.2018.8639121"},{"key":"ref119","article-title":"Low-latency broadband analog aggregation for federated edge learning","author":"zhu","year":"2018","journal-title":"arXiv 1812 11494"},{"key":"ref114","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2017.2761740"},{"key":"ref113","doi-asserted-by":"publisher","DOI":"10.1145\/2973750.2973777"},{"key":"ref80","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM.2019.8737614"},{"key":"ref116","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01234-2_48"},{"key":"ref115","article-title":"pCAMP: Performance comparison of machine learning packages on the edges","author":"zhang","year":"2018","journal-title":"Proceedings of USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18)"},{"key":"ref89","article-title":"Adaptive neural networks for efficient inference","author":"bolukbasi","year":"2017","journal-title":"arXiv 1702 07811"},{"key":"ref120","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.351"},{"key":"ref121","doi-asserted-by":"publisher","DOI":"10.1109\/SEC.2018.00058"},{"key":"ref122","doi-asserted-by":"publisher","DOI":"10.1109\/MCOM.2018.1701148"},{"key":"ref85","doi-asserted-by":"publisher","DOI":"10.23919\/DATE.2017.7927211"},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2018.2858384"},{"key":"ref87","doi-asserted-by":"publisher","DOI":"10.1109\/ICPR.2016.7900006"},{"key":"ref88","doi-asserted-by":"publisher","DOI":"10.1109\/ICDCS.2017.226"}],"container-title":["Proceedings of the IEEE"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/5\/8789751\/08736011.pdf?arnumber=8736011","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,1,12]],"date-time":"2022-01-12T16:53:54Z","timestamp":1642006434000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/8736011\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,8]]},"references-count":122,"journal-issue":{"issue":"8"},"URL":"https:\/\/doi.org\/10.1109\/jproc.2019.2918951","relation":{},"ISSN":["0018-9219","1558-2256"],"issn-type":[{"value":"0018-9219","type":"print"},{"value":"1558-2256","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,8]]}}}