{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,4]],"date-time":"2026-03-04T17:00:58Z","timestamp":1772643658606,"version":"3.50.1"},"reference-count":44,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"12","license":[{"start":{"date-parts":[[2022,12,1]],"date-time":"2022-12-01T00:00:00Z","timestamp":1669852800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Comput."],"published-print":{"date-parts":[[2022,12,1]]},"DOI":"10.1109\/tc.2022.3202747","type":"journal-article","created":{"date-parts":[[2022,9,22]],"date-time":"2022-09-22T23:04:53Z","timestamp":1663887893000},"page":"3179-3190","source":"Crossref","is-referenced-by-count":19,"title":["LNS-Madam: Low-Precision Training in Logarithmic Number System Using Multiplicative Weight Update"],"prefix":"10.1109","volume":"71","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-5726-6040","authenticated-orcid":false,"given":"Jiawei","family":"Zhao","sequence":"first","affiliation":[{"name":"Caltech, Pasadena, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5045-1964","authenticated-orcid":false,"given":"Steve","family":"Dai","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"given":"Rangharajan","family":"Venkatesan","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9997-3141","authenticated-orcid":false,"given":"Brian","family":"Zimmer","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4452-6464","authenticated-orcid":false,"given":"Mustafa","family":"Ali","sequence":"additional","affiliation":[{"name":"Purdue University, West Lafayette, IN, USA"}]},{"given":"Ming-Yu","family":"Liu","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7584-3489","authenticated-orcid":false,"given":"Brucek","family":"Khailany","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4632-2876","authenticated-orcid":false,"given":"William J.","family":"Dally","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, CA, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6974-6797","authenticated-orcid":false,"given":"Anima","family":"Anandkumar","sequence":"additional","affiliation":[{"name":"Caltech, Pasadena, CA, USA"}]}],"member":"263","reference":[{"key":"ref1","first-page":"1737","article-title":"Deep learning with limited numerical precision","volume-title":"Proc. 32nd Int. Conf. Mach. Learn.","author":"Gupta"},{"key":"ref2","first-page":"7686","article-title":"Training deep neural networks with 8-bit floating point numbers","volume-title":"Proc. Adv. Neural Informat. Process. Syst. 31: Annu. Conf. Neural Inf. Process. Syst.","author":"Wang"},{"key":"ref3","article-title":"The lottery ticket hypothesis: Finding sparse, trainable neural networks","volume-title":"Proc. 7th Int. Conf. Learn. Representations","author":"Frankle"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.7554\/eLife.10778"},{"key":"ref5","article-title":"Convolutional neural networks using logarithmic data representation","author":"Miyashita","year":"2016"},{"key":"ref6","article-title":"Ultra-low precision 4-bit training of deep neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst. 33: Annu. Conf. Neural Informat. Process. Syst.","author":"Sun"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3240765.3240803"},{"key":"ref8","article-title":"Learning compositional functions via multiplicative weight updates","volume-title":"Proc. Adv. Neural Inf. Process. Syst. 33: Annu. Conf. Neural Inf. Process. Syst.","author":"Bernstein"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/TEC.1962.5219391"},{"key":"ref10","article-title":"Estimating or propagating gradients through stochastic neurons for conditional computation","author":"Bengio","year":"2013"},{"key":"ref11","article-title":"Neural gradients are near-lognormal: Improved quantized and sparse training","volume-title":"Proc. 9th Int. Conf. Learn. Representations","author":"Chmiel"},{"key":"ref12","article-title":"signSGD with majority vote is communication efficient and fault tolerant","volume-title":"Proc. 7th Int. Conf. Learn. Representations","author":"Bernstein"},{"key":"ref13","first-page":"560","article-title":"signSGD: Compressed optimisation for non-convex problems","volume-title":"Proc. 35th Int. Conf. Mach. Learn.","author":"Bernstein"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD45719.2019.8942127"},{"key":"ref15","article-title":"Integer quantization for deep learning inference: Principles and empirical evaluation","author":"Wu","year":"2020"},{"key":"ref16","article-title":"A statistical framework for low-bitwidth training of deep neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst. 33: Annu. Conf. Neural Inf. Process. Syst.","author":"Chen"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.00204"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/5.52214"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053015"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2017.7953288"},{"key":"ref21","article-title":"Efficient large-scale language model training on GPU clusters","author":"Narayanan","year":"2021","journal-title":"Proc. Int. Conf. High Perform. Comput., Netw., Storage Anal."},{"key":"ref22","article-title":"DoReFa-Net: Training low bitwidth convolutional neural networks with low bitwidth gradients","author":"Zhou","year":"2016"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-46493-0_32"},{"key":"ref24","article-title":"Incremental network quantization: Towards lossless CNNs with low-precision weights","volume-title":"Proc. 5th Int. Conf. Learn. Representations","author":"Zhou"},{"key":"ref25","first-page":"2704","article-title":"Quantization and training of neural networks for efficient integer-arithmetic-only inference","volume-title":"Proc. IEEE Conf. Comput. Vis. Pattern Recognit.","author":"Jacob"},{"key":"ref26","first-page":"5151","article-title":"Scalable methods for 8-bit training of neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst. 31: Annu. Conf. Neural Inf. Process. Syst.","author":"Banner"},{"key":"ref27","article-title":"Per-tensor fixed-point quantization of the back-propagation algorithm","volume-title":"Proc. 7th Int. Conf. Learn. Representations","author":"Sakr"},{"key":"ref28","article-title":"Training and inference with integers in deep neural networks","volume-title":"Proc. 6th Int. Conf. Learn. Representations","author":"Wu"},{"key":"ref29","first-page":"4901","article-title":"Hybrid 8-bit floating point (HFP8) training and inference for deep neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst. 32: Annu. Conf. Neural Inf. Process. Syst.","author":"Sun"},{"key":"ref30","article-title":"High-accuracy low-precision training","author":"Sa","year":"2018"},{"key":"ref31","first-page":"706","article-title":"Cambricon-Q: A hybrid architecture for efficient training","volume-title":"Proc. ACM\/IEEE 48th Annu. Int. Symp. Comput. Architecture.","author":"Zhao"},{"key":"ref32","article-title":"Rethinking floating point for deep learning","author":"Johnson","year":"2018"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/ARITH48897.2020.00013"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2018.2857262"},{"key":"ref35","first-page":"1","article-title":"Approximate integer and floating-point dividers with near-zero error bias","volume-title":"Proc. ACM\/IEEE 56th Des. Automat. Conf.","author":"Saadat"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.23919\/DATE48585.2020.9116315"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1006\/inco.1996.2612"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1006\/jcss.1997.1504"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref40","first-page":"4171","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","volume-title":"Proc. Conf. North Amer. Chapter Assoc. Comput. Linguistics: Hum. Lang. Technol.","author":"Devlin"},{"key":"ref41","article-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2009"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref43","first-page":"2383","article-title":"SQuAD: 100,000+ questions for machine comprehension of text","volume-title":"Proc. Conf. Empir. Methods Natural Lang. Process.","author":"Rajpurkar"},{"key":"ref44","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/W18-5446","article-title":"GLUE: A multi-task benchmark and analysis platform for natural language understanding","volume-title":"Proc. 7th Int. Conf. Learn. Representations","author":"Wang"}],"container-title":["IEEE Transactions on Computers"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/12\/9953587\/09900267.pdf?arnumber=9900267","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,5,16]],"date-time":"2024-05-16T05:09:54Z","timestamp":1715836194000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9900267\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,12,1]]},"references-count":44,"journal-issue":{"issue":"12"},"URL":"https:\/\/doi.org\/10.1109\/tc.2022.3202747","relation":{},"ISSN":["0018-9340","1557-9956","2326-3814"],"issn-type":[{"value":"0018-9340","type":"print"},{"value":"1557-9956","type":"electronic"},{"value":"2326-3814","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,12,1]]}}}