{“状态”:“确定”,“消息类型”:“工作”,“信息版本”:“1.0.0”,“邮件”:{“索引”:{“日期-部件”:[[2024,6,20]],“日期-时间”:“2024-06-20T08:36:07Z”,“时间戳”:1718872567920},“参考-计数”:35,“出版商”:“Springer Science and Business Media LLC”我“:“2021-10-01T00:00:00Z”,“timestamp”:163304640000},“content-version”:“tdm”,“delay-in-days”:0,“URL”:“https:\\/www.springer.com/tdm”},{“start”:{“date-parts”:[[2021,10,1]],“date-time”:“2021-10-01T00:00:00Z”,“timetamp”:633046400000},”content-versiion“:”vor“,”delay-in-days“:0,”URL“https://www.springer.com\/tdm“}”,“出资人”:[{“名称”:“Hiroshi Fujiwara Technion网络安全研究中心”,“奖项”:[“N\/A”]}],“内容域”:{“域”:[”link.springer.com“],“crossmark-restriction”:false},“short-container-title”:[)Mach Learn“],”published-print“:{”date-parts“:[2021,12]]},”DOI“:”10.1007\/s10994-021-06053-z“,”type“:”journal-article“,”created“:{”date-ports“”:[2021,10,1]],“日期-时间”:“2021-10-01T15:19:30Z”,“时间戳”:1633101570000},“页面”:“3245-3262”,“更新策略”:“http:\/\/dx.doi.org/10.1007\/springer_crossmark_policy”,“源”:“Crossref”,“被计数引用”:49,“标题”:[“损失感知训练后量化”],“前缀”:“10.1007”,“卷”:“110”,“作者”:[{“给定”:“Yury”,“家族”:“Nahshan”,“序列”:“first”,“affiliation”:[]},{“given”:“Brian”,“family”:“Chmiel”,“sequence”:“additional”,“affiliation”:[]},{“given”:“Chaim”,“family”:“Baskin”,“sequence”:“additional”,“affiliation”:[]},{“ORCID”:“http:\/\/ORCID.org/0000-0002-5400-9321”,“authenticated ORCID”:false,“given”:“Evgenii”,“family”:“Zheltonozkii”,“sequence”:“additional”,“附属关系”:[]},{“given”:“Ron”,“family”:“Banner”,“sequence”:“additional”,“affiliation”:[]},{“给定”:“Alex M.”,“家族”:“Bronstein”,”sequence“:”additional“,”affiliance“:[]{”given“:”Avi“,”family“:”Mendelson“,”se序列“:”additional:[{“key”:“6053_CR1”,“非结构化”:“Banner,R.,Nahshan,Y.,Hoffer,E.,Soudry,D.(2018)快速部署卷积网络的训练后4位量化。arXiv预印本arXiv:181005723http://\/arXiv.org\/abs\/1810.05723”},{“key”:“6053_CR2”,“unstructured”:“Baskin,C.,Liss,N.,Chai,Y.、Zheltonzhkii,E.,Schwartz,E.,Girayes,R.、Mendelson,A.,Bronstein,A。M.(2018)尼斯:神经网络量化的噪声注入和钳制估计。arXiv预印本arXiv:181000162http:\/\/arXiv.org\/abs\/1810.00162“},{“问题”:“6”,“密钥”:“6053_CR3”,“doi-asserted-by”:“发布者”,“首页”:“759”,“doi”:“10.1109\/TIT.1972.1054906”,“卷”:“18”,“作者”:“T Berger”,“年份”:“1972”,“非结构化”:“Berger,T.(1972).最佳量化器和置换码。IEEE Transactions on Information Theory,18(6),759\u2013765。“,“journal-title”:“IEEE Translations on Information Theory”},{“key”:“6053_CR4”,“unstructured”:“Bethge,J.,Yang,H.,Bornstein,M.,Meinel,C.,(2019)回到简单:如何从头开始训练准确的bnns?arXiv预印本arXiv:190608637http://arXiv.org\/abs\/1906.08637”}、{“key”:6053_CR5“,“非结构化”:”Brent,R.P.(2013)《无导数最小化算法》。Courier公司。“},{”key“:”6053_CR6“,”doi-asserted-by“:”crossref“,”unstructured“:”Fang,J.,Shafie,A.,Abdel-Aziz,H.,Thorsley,D.,Georgiadis,G.,Hassoun,J.(2020)通过分段线性近似对深层神经网络进行近无损训练后量化。arXiv预印本arXiv:200200104https:\//arXiv.org\/abs\/2002.00104“,”doi“:”10.1007\/978-3-030-58536-5_5“},{“key”:“6053_CR7”,“doi-asserted-by”:“crossref”,“unstructured”:“Faraone,J.,Fraser,N.,Blott,M.,Leong,P.H.(2018)Syq:高效深层神经网络的学习对称量化。In:IEEE计算机视觉和模式识别会议(CVPR),http://\/openaccess.thevf.com/content_cvpr_2018\/html\/Faraone_SYQ_Learning_Symetric_cvpr_2018_paper.html“,“DOI”:“10.1109\/cvpr.2018.00452”},{“key”:“6053_CR8”,“unstructured”:“Fawzi,A.,Moosavi-Dezfouli,S.M.,Frossard,P.,(2016)分类器的鲁棒性:从对抗噪声到随机噪声。arXiv-print arXiv:160808967https:\/\/arXiv.org\/abs\/1608.08967“},{“key”:“6053_CR9”,“unstructured”:“Finkelstein,A.,Almog,U.,Grobman,M.(2019)与量化偏差作斗争。arXiv-Print arX iv:190603193http:\/\/arXiv.org\/abs\/1906.03193”},}“issue”:“7”,“key“:”6053_CR10“,”“doi-asserted-by”:“publisher”,“首页”:“632”,“doi”:“10.1016\/j.cagd.2005.06.005“,”volume“:”22“,”author“:”R Goldman“,”year“:”2005“,”unstructured“:”Goldman,R.(2005)。隐式曲线和曲面的曲率公式。计算机辅助几何设计,22(7),632\u2013658。https:\/\/doi.org\/10.1016\/j.cagd.2005.06.005.“,“journal-title”:“计算机辅助几何设计”},{“key”:“6053_CR11”,“doi-asserted-by”:“crossref”,“unstructured”:“龚,j.,Shen,H.,Zhang,G.,Liu,X.,Li,S.,Jin,G.、Maheshwari,N.,Fomenko,E.,Segal,E.(2018)基于intelcaffe的卷积神经网络高效8位低精度推理。摘自:美国纽约州纽约市ACM,第一届可复制质量效率系统协同设计Pareto-Efficient Deep Learning竞赛论文集,ReQuEST\u201918,10.1145\/3229762.3229763,http://doi.ACM.org\/10.1145\/322962.3229763“,“doi”:“10.1145\/3229762.32297633”},{“key”:“6053_CR12”,“doi-asserted-by”:“crossref”,“unstructured”:“Gong,R.,Liu,X.,Jiang,S.,Li,T.,Hu,P.,Lin,J.,Yu,F.,Yan,J.(2019)可微分软量化:桥接全精度和低比特神经网络。摘自:IEEE计算机视觉国际会议(ICCV),http://\/openaccess.thevf.com/content_ICCV_2019\/html\/Gong_Differentiable_Soft_Quantization_Bridging_Full-Precision_and_Low-Bit_Neural_Networks_ICCCV_2019_paper.html“,”DOI“:”10.1109\/ICCV.2019.00495“},{“key”:“6053_CR13”,“DOI-asserted-by”:“crossref”,“unstructured”:“He,X.,Liao,L.,Zhang,H.,Nie,L.、Hu,X.Chua,T.S.(2017)神经协同过滤。摘自:《第26届万维网国际会议论文集》,国际万维网会议指导委员会,瑞士日内瓦共和国和州,WWW\u201917,pp 173\u2013182,10.1145\/3038912.3052569,https:\//doi.org\/10.1145\/3038912.3052569“,”doi“:”10.1145\/3038.912.3052569“},{“key”:“6053_CR14”,“unstructured”:“Hou,L.,Kwok,J.T.(2018)深度网络的损失感知权重量化。https:\/\/arxiv.org\/abs\/1802.08635深层网络的损失感知二值化。arXiv预印本arXiv:16101600 https:\/\/arXiv.org/abs\/161011600“},{“问题”:“1”,“密钥”:“6053_CR16”,“第一页”:“6869”,“卷”:“18”,“作者”:“I Hubara”,“年份”:“2017”,“非结构化”:“Hubara,I.,Courbareaux,M.,Soudry,D.,El Yaniv,R.,&Bengio,Y.(2017)量化神经网络:训练具有低精度权重和激活的神经网络。J Mach Learn Res,18(1),6869\u20136898.“,“journal-title”:“J Mach Learn Res”},{“key”:“6053_CR17”,“unstructured”:“Jin,Q.,Yang,L.,Liao,Z.(2019)迈向神经网络量化的高效训练。arXiv预印本arXiv:191210207https:\/\/arXiv.org\/abs\/1912.10207”}、{“key”:”6053_CR18“非结构化”:“Kim,H.,Kim,K.,Kim。J.(2020)Binaryduo:通过耦合二进制激活来减少二进制激活网络中的梯度失配。arXiv预印本arXiv:200206517https:\/\/arXiv.org\/abs\/2002.06517“},{“key”:“6053_CR19”,“unstructured”:“Kravchik,E.,Yang,F.,Kisilev,P.,Choukroun,Y.(2019)用于有效推理的神经网络低比特量化。In:IEEE国际计算机视觉会议(ICCV)研讨会,http://\/openaccess.thevf.com/content_ICCVW_2019\/html\/CEFRL\/Kravchik_Low-bit_Quantization_of_Neural_Networks_for_Efficient_Inference_ICCCVW_2019 _paper.html“},{“key”:“6053_CR20”,“unstructured”:“Krishnamoorthi,R.(2018)量化深度卷积网络以进行有效推理:白皮书。arXiv预印本arXiv:180608342http:\/\/arXiv.org\/abs\/1806.08342“},{“key”:“6053_CR21”,“unstructured”:“Lee,J.H.,Ha,S.,Choi,S,Lee,W.J.,Lee and S.(2018)深度神经网络快速部署的量化。arXiv-print arXiv:18100548http:\//arXiv.org \/abs\/1810.05488”},}“key”:Lin,D.D.,Talathi,S.S.,Annapureddy,V.S.(2015)深卷积网络的定点量化。arXiv-print arXiv:151106393https:\/\/arXiv.org\/abs\/1511.06393“},{“key”:“6053_CR23”,“doi-asserted-by”:“crossref”,“unstructured”:“Liu,C.,Ding,W.,Xia,Hu,Y.,Zhang,B.,Liu,J.,Zhuang,B.,Guo,G.(2019)改进的二进制卷积网络,用于提高1位DCNN的性能。摘自:《第二十届国际人工智能联合会议论文集》,IJCAI-19,国际人工智能组织联合会议,第854\u2013860页,第10.24963\/IJCAI.2019\/120页,https:\/\/doi.org\/10.24963\/IJCAI.2019\/120“,“doi”:“10.24963\/IJCAI.2019\/120”},{“issue”:“2”,“key”:“6053_CR24”,“doi由“:”出版商“,”第一页“:”129“,”doi“:”10.1109\/TIT.1982.1056489“,”卷“:”28“,”作者“:”S Lloyd“,”年份“:”1982“,”非结构化“:”Lloyd,S.(1982)。PCM中的最小二乘量化。IEEE Transactions on Information Theory,28(2),129\u2013137.“,“journal-title”:“IEEE Translations on Information Theory”},{“issue”:“1”,“key”:“6053_CR25”,“doi-asserted-by”:“publisher”,“first page”:《7》,“doi”:“10.1109\/TIT.1960.1057548”,“volume”:第6卷,“author”:“J Max”,“year”:“1960”,“unstructured”:“Max,J.(1960)量化以实现最小失真。IRE Transactions on Information Theory,6(1),7\u201312.“,“journal title”:“IRE Translations on Information Theory”},{“key”:“6053_CR26”,“unstructured”:“Mellempudi,N.,Kundu,A.,Mudigere,D.,Das,D.,Kaul,B.,Dubey,P.(2017)具有精细量化的三值神经网络。arXiv预印本arXiv:170501462http:\/\/arXiv.org\/abs\/1705.01462“},{“key”:“6053_CR27”,“unstructured”:“Migacz,S.(2017)使用张量的8位推理。http://\/on-demand.gputechconf.com\/gtc\/2017\/presentationation\/s7310-8-bit-inference-with-tensorrt.pdf”}“:”Nagel,M.,Baalen,M.v.,Blankevort,T.,Welling,M.(2019)通过权重均衡和偏差校正实现无数据量化。摘自:IEEE计算机视觉国际会议论文集,第1325\u20131334页,http://\/openaccess.thevf.com/content_ICCV_2019\/html\/Nagel_Data-Free_Quantization_Through_Weight_Equalization_and_Bias_Correction_ICCV_2 019_paper.html“,”DOI“:”10.1109\/ICCCV.2019.00141“},{“key”:“6053_CR29”,“DOI-asserted-by”:“crossref”,“unstructured”:“Nayak,P.,Zhang,D.,Chai,S.(2019)深度神经网络的比特有效量化。arXiv预印本arXiv:191004877http:\/\/arXiv.org\/abs\/1910.04877“,”DOI“:”10.1109\/EMC2-NIPS5302019.00020“},{“key”:“6053_CR30”,“DOI-asserted-by”:“publisher”,“DOI”:“10.1016\/j.patrec.2019.03.026”,“author”:“H Peng”,“year”:“2019”,“unstructured”:“Peng,H.,&Chen,S.(2019).Bdnn:用于快速目标检测的二进制卷积神经网络。模式识别字母。https:\/\/doi.org\/10.1016\/j.patrec.2019.03.026.“,”journal-title“:”Pattern Recognition Letters“},{”issue“:”2“,”key“:”6053_CR31“,”doi-asserted-by“:”publisher“,“first page”:“155”,“doi”:“10.1093\/comjnl\/7.2.155”、“volume“7”、“author”:“MJ Powell”,“year”:“1964”,“unstructured”:“Powell,M.j.(1964)一种不计算导数而求多变量函数最小值的有效方法。《计算机杂志》,7(2),155.“,”Journal-title“:”The Computer Journal“},{”key“:”6053_CR32“,”doi-asserted-by“:”crossref“,”unstructured“:”Yang,J.,Shen,X.,Xing,J.、Tian,X.、Li,H.、Deng,B.、Huang,J.和Hua,X.s.(2019)量化网络。In:IEEE Conference on Computer Vision and Pattern Recognition(CVPR),http://\/openaccess.thevf.com/content_CVPR_2019\/html\/Yang_Quantization_Networks_CVPR_2 019_paper.html“,”DOI“:”10.1109\/CVPR.2019.00748“},{“key”:“6053_CR33”,“DOI-asserted-by”:“crossref”,“unstructured”:“Zhang,D.,Yang,J.,Ye,D.,Hua,G.(2018)Lq-nets:用于高精度和紧凑的深层神经网络的学习量化。在:欧洲计算机视觉会议(ECCV),http://\/openaccess.thevf.com/content_ECCV_2018\/html\/Dongqing_Zhang_Optimized_Quantization_for_ECCV_2 018_paper.html,“DOI”:“10.1007\/978-3-030-01237-3_23”},{“key”:“6053_CR34”,“unstructured”:“Zhao,R.,Hu,Y.,Dotzel,J.,De\u00a0Sa,C.,Zhang,Z.(2019)使用离群信道分裂改进神经网络量化。arXiv预打印arXiv:190109504https:\/\/arXiv.org\/abs\/1901.09504“},{“key”:“6053_CR35”,“doi-asserted-by”:“crossref”,“unstructured”:“Zhou,Y.,Moosavi-Dezfouli,S.M.,Cheung,N.M.,Frossard,P.(2017)深度神经网络的自适应量化。arXiv-print arXiv:171201048https:\//arXiv.org/abs\/1712.01048”,“doi”:“10.1609\/aaai.v32i1.11623“}],“container-title”:[“机器学习”],“原始标题”:[],“language”:“en”,“link”:[{“URL”:“https:\/\/link.springer.com/content\/pdf\/10.1007\/s10994-021-06053-z.pdf”,“content-type”:“application\/pdf”、“content-version”:“vor”,“intended-application”:“text-mining”},{“URL”:“https:\/\/link.springer.com/article\/10.1007\/s10994-021-06053-z\/fulltext.html”,“content-type”:“text\/html”,“内容-版本”:“vor”,“intended-application”:“文本-分钟”},{“URL”:“http:\/\-link.springer.com\/content\/pdf\/10007\/s10994-021-0653-z.pdf”,“content-type“:”application\/pdf“,”content-version“:”vor“,”intended d-应用程序“:“相似性检查”}],“存放”:{“日期部分”:[[2022,10,1]],“日期时间”:“2022-10-01T02:08:08Z”,“时间戳”:1664590088000},“分数”:1,“资源”:{primary“:{”URL“:”https:\/\/link.springer.com\/10.1007\/s10994-021-06053-z“}”,“副标题”:[],“短标题”:[],“已发布”:}“日期部件”:[2021,10,1]]},“references-count”:35,“journal-issue”:{“issue”:“11-12”,“published-print”:{“date-parts”:[2021,12]]}},“alternative-id”:[“6053”],“URL”:“http://\/dx.doi.org\/10.1007\/s10994-021-06053-z”,“relationship”:{},”ISSN“:[“0885-6125”,“1573-0565”],”ISSN-type“:[{“value”:“0885-6 125”,”type“”:“print”},{“value”:“1573-00565”,“type”:“电子”}],“主题”:[],“发布”:{“日期部分”:[2021,10,1]]},“断言”:[{“值”:“2020年11月29日”,“订单”:1,“名称”:“已收到”,“标签”:“收到”,”组“:{“name”:“Article History”,“label”:“文章历史”}},{“value”:“2021年6月10日”,”订单“:2,”名称“:”修订“,”标签“:“修订”,“组“:ed“,”group“:{”name“:“ArticleHistory”,“label”:“ArticleHistory”}},{“value”:“2021年10月1日”,“order”:4,“name”:“first_online”,“label”:“first online”,“group”:{“name”:“ArticleHistory”,“label”:“ArticleHistory”},{“value”:“此内容已向所有人开放。”,“name”:“free”,“label”:“free to read”}