{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,27]],"date-time":"2025-10-27T18:09:04Z","timestamp":1761588544225,"version":"build-2065373602"},"reference-count":61,"publisher":"IBM","issue":"6","funder":[{"DOI":"10.13039\/100002418","name":"Intel Corporation","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100002418","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100004358","name":"Samsung","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100004358","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100017567","name":"Apple","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100017567","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100006785","name":"Google","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100006785","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100016822","name":"Seagate Technology","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100016822","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IBM J. Res. &amp; Dev."],"published-print":{"date-parts":[[2019,11,1]]},"DOI":"10.1147\/jrd.2019.2942284","type":"journal-article","created":{"date-parts":[[2019,9,18]],"date-time":"2019-09-18T16:08:35Z","timestamp":1568822915000},"page":"6:1-6:14","source":"Crossref","is-referenced-by-count":5,"title":["Co-design of deep neural nets and neural net accelerators for embedded vision applications"],"prefix":"10.1147","volume":"63","author":[{"given":"A.","family":"Amid","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6515-755X","authenticated-orcid":false,"given":"K.","family":"Kwon","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"A.","family":"Gholami","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"B.","family":"Wu","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"K.","family":"Asanovi\u0107","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3868-8501","authenticated-orcid":false,"given":"K.","family":"Keutzer","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"3082","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2019.8662447"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2019.8662302"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2017.29"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1145\/2749469.2750389"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/MC.1982.1653825"},{"key":"ref30","first-page":"256","article-title":"Systolic arrays (for VLSI)","volume":"1","author":"kung","year":"1979","journal-title":"Proc Sparse Matrix Proc"},{"key":"ref37","first-page":"222","article-title":"An always-on 3.8 uJ\/ 86% CIFAR-10 mixed-signal binary CNN processor with all memory on chip in 28nm CMOS","author":"bankman","year":"2018","journal-title":"Proc IEEE Int Solid-State Circuits Conf"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2017.7870349"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3079856.3080254"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001163"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1145\/3125502.3125606"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1109\/40.946678"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/1562764.1562783"},{"article-title":"The landscape of parallel computing research: A view from Berkeley","year":"2006","author":"asanovic","key":"ref27"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/1953611.1953620"},{"article-title":"Learning multiple layers of features from tiny images","year":"2009","author":"krizhevsky","key":"ref2"},{"year":"2010","author":"lecun","key":"ref1"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2018.8310262"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2017.7870353"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2018.8310261"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2007.445"},{"year":"2017","key":"ref23"},{"journal-title":"Computer Architecture A Quantitative Approach","year":"2011","author":"hennessy","key":"ref26"},{"year":"2018","key":"ref25"},{"article-title":"Very deep convolutional networks for large-scale image recognition","year":"2014","author":"simonyan","key":"ref50"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2018.8310176"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1145\/3195970.3199849"},{"article-title":"DNN dataflow choice is overrated","year":"2018","author":"yang","key":"ref58"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00716"},{"article-title":"Mobilenets: Efficient convolutional neural networks for mobile vision applications","year":"2017","author":"howard","key":"ref56"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00951"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2015.7298594"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2019.8662396"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2012.6248074"},{"year":"2017","key":"ref11","article-title":"Evaluating efficiency of several types of convolutions"},{"key":"ref40","first-page":"138","article-title":"A 65nm 0.39-to-140.3 TOPS\/W 1-to-12b unified neural-network processor using block-circulant-enabled transpose-domain acceleration with 8.1&#x00D7; higher TOPS\/mm2 and 6T HBST-TRAM-based 2D data-reuse architecture","author":"yue","year":"2019","journal-title":"Proc IEEE Int Solid-State Circuits Conf"},{"key":"ref12","first-page":"409","article-title":"SkipNet: Learning dynamic routing in convolutional networks","author":"wang","year":"2018","journal-title":"Proc Eur Conf Comput Vision"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001177"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW.2018.00215"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.243"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/1498765.1498785"},{"article-title":"Execution-cache-memory performance model: Introduction and validation","year":"2015","author":"hufmann","key":"ref17"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2014.6757323"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1145\/3079856.3080246"},{"article-title":"EfficientNet-EdgeTPU: Creating accelerator-optimized neural networks with AutoML","year":"2019","author":"gupta","key":"ref4"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.351"},{"key":"ref6","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","author":"krizhevsky","year":"2012","journal-title":"Proc 25th Int Conf Neural Inf Process Syst"},{"article-title":"SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5 MB model size","year":"2016","author":"iandola","key":"ref5"},{"key":"ref8","first-page":"740","article-title":"Microsoft Coco: Common objects in context","author":"lin","year":"2014","journal-title":"Proc Eur Conf Comput Vis"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2017.7870350"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.350"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/VLSIC.2018.8502333"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/VLSIC.2018.8502404"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/VLSIC.2018.8502276"},{"key":"ref47","first-page":"37","article-title":"An ultra-high energy-efficient reconfigurable processor for deep neural networks with binary\/ternary weights in 28nm CMOS","author":"yin","year":"2018","journal-title":"Proc Symp VLSI Circuits"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.23919\/VLSIC.2017.8008534"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2019.8662476"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.23919\/VLSIC.2017.8008533"},{"key":"ref43","first-page":"30c","article-title":"A 3.43 TOPS\/W 48.9 pJ\/pixel 50.1 nJ\/classification 512 analog neuron sparse coding neural network with on-chip learning and classification in 40nm CMOS","author":"buhler","year":"2017","journal-title":"Proc Symp VLSI Circuits"}],"container-title":["IBM Journal of Research and Development"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/5288520\/8894910\/08843941.pdf?arnumber=8843941","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,27]],"date-time":"2025-10-27T18:03:35Z","timestamp":1761588215000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/8843941\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,11,1]]},"references-count":61,"journal-issue":{"issue":"6"},"URL":"https:\/\/doi.org\/10.1147\/jrd.2019.2942284","relation":{},"ISSN":["0018-8646","0018-8646"],"issn-type":[{"type":"print","value":"0018-8646"},{"type":"electronic","value":"0018-8646"}],"subject":[],"published":{"date-parts":[[2019,11,1]]}}}