{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,6]],"date-time":"2025-12-06T16:47:24Z","timestamp":1765039644330,"version":"3.37.3"},"reference-count":51,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2022,1,1]],"date-time":"2022-01-01T00:00:00Z","timestamp":1640995200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2022,1,1]],"date-time":"2022-01-01T00:00:00Z","timestamp":1640995200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2022,1,1]],"date-time":"2022-01-01T00:00:00Z","timestamp":1640995200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100012166","name":"National Key Research and Development Program of China","doi-asserted-by":"publisher","award":["2017YFA0700902","2017YFA0700903"],"award-info":[{"award-number":["2017YFA0700902","2017YFA0700903"]}],"id":[{"id":"10.13039\/501100012166","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"NSF of China","doi-asserted-by":"publisher","award":["61925208","61906179","62102399","61732020","U19B2019"],"award-info":[{"award-number":["61925208","61906179","62102399","61732020","U19B2019"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002367","name":"Strategic Priority Research Program of Chinese Academy of Science","doi-asserted-by":"publisher","award":["XDB32050200"],"award-info":[{"award-number":["XDB32050200"]}],"id":[{"id":"10.13039\/501100002367","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Beijing Academy of Artificial Intelligence (BAAI) and the Beijing Nova Program of Science and Technology","award":["Z191100001119093"],"award-info":[{"award-number":["Z191100001119093"]}]},{"DOI":"10.13039\/501100002367","name":"CAS Project for Young Scientists in Basic Research","doi-asserted-by":"publisher","award":["YSBR-029"],"award-info":[{"award-number":["YSBR-029"]}],"id":[{"id":"10.13039\/501100002367","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100012492","name":"Youth Innovation Promotion Association","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100012492","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. on Image Process."],"published-print":{"date-parts":[[2022]]},"DOI":"10.1109\/tip.2022.3216776","type":"journal-article","created":{"date-parts":[[2022,11,4]],"date-time":"2022-11-04T01:47:40Z","timestamp":1667526460000},"page":"7006-7019","source":"Crossref","is-referenced-by-count":11,"title":["Rethinking the Importance of Quantization Bias, Toward Full Low-Bit Training"],"prefix":"10.1109","volume":"31","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-9540-6086","authenticated-orcid":false,"given":"Chang","family":"Liu","sequence":"first","affiliation":[{"name":"SKL of Computer Architecture, Institute of Computing Technology, CAS, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xishan","family":"Zhang","sequence":"additional","affiliation":[{"name":"SKL of Computer Architecture, Institute of Computing Technology, CAS, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8691-8549","authenticated-orcid":false,"given":"Rui","family":"Zhang","sequence":"additional","affiliation":[{"name":"SKL of Computer Architecture, Institute of Computing Technology, CAS, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8877-9052","authenticated-orcid":false,"given":"Ling","family":"Li","sequence":"additional","affiliation":[{"name":"University of Chinese Academy of Sciences, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Shiyi","family":"Zhou","sequence":"additional","affiliation":[{"name":"Cambricon Technologies, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Di","family":"Huang","sequence":"additional","affiliation":[{"name":"SKL of Computer Architecture, Institute of Computing Technology, CAS, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhen","family":"Li","sequence":"additional","affiliation":[{"name":"Cambricon Technologies, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7603-4210","authenticated-orcid":false,"given":"Zidong","family":"Du","sequence":"additional","affiliation":[{"name":"SKL of Computer Architecture, Institute of Computing Technology, CAS, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Shaoli","family":"Liu","sequence":"additional","affiliation":[{"name":"Cambricon Technologies, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tianshi","family":"Chen","sequence":"additional","affiliation":[{"name":"Cambricon Technologies, Beijing, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"doi-asserted-by":"publisher","key":"ref39","DOI":"10.1109\/CVPR.2016.308"},{"doi-asserted-by":"publisher","key":"ref38","DOI":"10.1109\/CVPR.2015.7298594"},{"key":"ref33","first-page":"7950","article-title":"Post training 4-bit quantization of convolutional networks for rapid-deployment","author":"ron","year":"2019","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref32","first-page":"91","article-title":"Faster R-CNN: Towards real-time object detection with region proposal networks","author":"ren","year":"2015","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"doi-asserted-by":"publisher","key":"ref31","DOI":"10.1016\/j.patcog.2020.107281"},{"doi-asserted-by":"publisher","key":"ref30","DOI":"10.1109\/CVPRW50498.2020.00357"},{"key":"ref37","first-page":"1796","article-title":"Ultra-low precision 4-bit training of deep neural networks","volume":"33","author":"sun","year":"2020","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref36","first-page":"4900","article-title":"Hybrid 8-bit floating point (HFP8) training and inference for deep neural networks","author":"sun","year":"2019","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref35","first-page":"1","article-title":"Very deep convolutional networks for large-scale image recognition","author":"simonyan","year":"2015","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref34","first-page":"1","article-title":"Per-tensor fixed-point quantization of the back-propagation algorithm","author":"sakr","year":"2019","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref28","article-title":"Mixed precision training with 8-bit floating point","author":"mellempudi","year":"2019","journal-title":"arXiv 1905 12334"},{"key":"ref27","article-title":"Microsoft COCO: Common objects in context","author":"lin","year":"2014","journal-title":"arXiv 1405 0312"},{"key":"ref29","first-page":"8026","article-title":"PyTorch: An imperative style, high-performance deep learning library","author":"paszke","year":"2019","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref2","first-page":"242","article-title":"A convergence theory for deep learning via over-parameterization","author":"allen-zhu","year":"2018","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref1","article-title":"TensorFlow: Large-scale machine learning on heterogeneous distributed systems","author":"abadi","year":"2016","journal-title":"arXiv 1603 04467"},{"doi-asserted-by":"publisher","key":"ref20","DOI":"10.1109\/CVPR.2016.90"},{"doi-asserted-by":"publisher","key":"ref22","DOI":"10.1145\/3316781.3317783"},{"key":"ref21","first-page":"4114","article-title":"Binarized neural networks","author":"hubara","year":"2016","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref24","first-page":"20415","article-title":"Position-based scaled gradient for model quantization and sparse training","author":"kim","year":"2020","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"doi-asserted-by":"publisher","key":"ref23","DOI":"10.18653\/v1\/E17-2068"},{"key":"ref26","first-page":"1097","article-title":"ImageNet classification with deep convolutional neural networks","author":"krizhevsky","year":"2012","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref25","first-page":"1740","article-title":"Flexpoint: An adaptive numerical format for efficient training of deep neural networks","author":"koster","year":"2017","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref50","article-title":"DoReFa-net: Training low bitwidth convolutional neural networks with low bitwidth gradients","author":"zhou","year":"2016","journal-title":"arXiv 1606 06160 [cs]"},{"doi-asserted-by":"publisher","key":"ref51","DOI":"10.1109\/CVPR42600.2020.00204"},{"doi-asserted-by":"publisher","key":"ref10","DOI":"10.1145\/2541940.2541967"},{"key":"ref11","first-page":"3123","article-title":"BinaryConnect: Training deep neural networks with binary weights during propagations","author":"courbariaux","year":"2015","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref40","article-title":"FAT: Learning low-bitwidth parametric representation via frequency-aware transformation","author":"tao","year":"2021","journal-title":"arXiv 2102 07444"},{"key":"ref12","first-page":"1","article-title":"Mixed precision training of convolutional neural networks using integer operations","author":"das","year":"2018","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"doi-asserted-by":"publisher","key":"ref13","DOI":"10.1109\/CVPR.2009.5206848"},{"doi-asserted-by":"publisher","key":"ref14","DOI":"10.1007\/978-3-030-58536-5_5"},{"key":"ref15","first-page":"3304","article-title":"Don’t waste your bits! squeeze activations and gradients for deep neural networks via tinyscript","author":"fu","year":"2020","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref16","first-page":"12127","article-title":"FracTrain: Fractionally squeezing bit savings both temporally and spatially for efficient DNN training","author":"fu","year":"2020","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref17","article-title":"A survey of quantization methods for efficient neural network inference","author":"gholami","year":"2021","journal-title":"arXiv 2103 13630"},{"doi-asserted-by":"publisher","key":"ref18","DOI":"10.1145\/1143844.1143891"},{"key":"ref19","first-page":"1","article-title":"Deep learning with limited numerical precision","author":"gupta","year":"2015","journal-title":"Proc Int Conf Mach Learn (ICML)"},{"key":"ref4","first-page":"5151","article-title":"Scalable methods for 8-bit training of neural networks","author":"banner","year":"2018","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref3","first-page":"5741","article-title":"Bayesian bits: Unifying quantization and pruning","author":"baalen","year":"2020","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"doi-asserted-by":"publisher","key":"ref6","DOI":"10.1137\/16M1080173"},{"doi-asserted-by":"publisher","key":"ref5","DOI":"10.1145\/3444943"},{"key":"ref8","first-page":"1","article-title":"Shifted and squeezed 8-bit floating point format for low-precision training of deep neural networks","author":"cambier","year":"2020","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"doi-asserted-by":"publisher","key":"ref7","DOI":"10.1109\/CVPR42600.2020.00242"},{"doi-asserted-by":"publisher","key":"ref49","DOI":"10.1609\/aaai.v35i4.16462"},{"key":"ref9","first-page":"883","article-title":"A statistical framework for low-bitwidth training of deep neural networks","author":"chen","year":"2020","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref46","article-title":"Achieving human parity in conversational speech recognition","author":"xiong","year":"2016","journal-title":"arXiv 1610 05256"},{"key":"ref45","article-title":"Google’s neural machine translation system: Bridging the gap between human and machine translation","author":"wu","year":"2016","journal-title":"arXiv 1609 08144"},{"doi-asserted-by":"publisher","key":"ref48","DOI":"10.1109\/CVPR42600.2020.00240"},{"doi-asserted-by":"publisher","key":"ref47","DOI":"10.1016\/j.neunet.2019.12.027"},{"key":"ref42","first-page":"6000","article-title":"Attention is all you need","author":"vaswani","year":"2017","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"},{"key":"ref41","first-page":"2","article-title":"Efficient inference with TensorRT","volume":"1","author":"vanholder","year":"2016","journal-title":"Proc GPU Technol Conf"},{"key":"ref44","first-page":"1","article-title":"Training and inference with integers in deep neural networks","author":"wu","year":"2018","journal-title":"Proc Int Conf Learn Represent (ICLR)"},{"key":"ref43","first-page":"7686","article-title":"Training deep neural networks with 8-bit floating point numbers","author":"wang","year":"2018","journal-title":"Proc Conf Workshop Neural Inf Process Syst (NeurIPS)"}],"container-title":["IEEE Transactions on Image Processing"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/83\/9626658\/09935812.pdf?arnumber=9935812","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,12,12]],"date-time":"2022-12-12T19:47:00Z","timestamp":1670874420000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9935812\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022]]},"references-count":51,"URL":"https:\/\/doi.org\/10.1109\/tip.2022.3216776","relation":{},"ISSN":["1057-7149","1941-0042"],"issn-type":[{"type":"print","value":"1057-7149"},{"type":"electronic","value":"1941-0042"}],"subject":[],"published":{"date-parts":[[2022]]}}}