{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,20]],"date-time":"2026-10-20T12:28:50Z","timestamp":1792499330805,"version":"4.5.0"},"reference-count":103,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"12","license":[{"start":{"date-parts":[[2023,12,1]],"date-time":"2023-12-01T00:00:00Z","timestamp":1701388800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2023,12,1]],"date-time":"2023-12-01T00:00:00Z","timestamp":1701388800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2023,12,1]],"date-time":"2023-12-01T00:00:00Z","timestamp":1701388800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Pattern Anal. Mach. Intell."],"published-print":{"date-parts":[[2023,12]]},"DOI":"10.1109\/tpami.2023.3321392","type":"journal-article","created":{"date-parts":[[2023,10,2]],"date-time":"2023-10-02T17:52:54Z","timestamp":1696269174000},"page":"15802-15819","source":"Crossref","is-referenced-by-count":5,"title":["Towards a Deeper Understanding of Global Covariance Pooling in Deep Learning: An Optimization Perspective"],"prefix":"10.1109","volume":"45","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3765-9787","authenticated-orcid":false,"given":"Qilong","family":"Wang","sequence":"first","affiliation":[{"name":"College of Intelligence and Computing, Tianjin University, Tianjin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-2727-820X","authenticated-orcid":false,"given":"Zhaolin","family":"Zhang","sequence":"additional","affiliation":[{"name":"College of Intelligence and Computing, Tianjin University, Tianjin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-5837-9926","authenticated-orcid":false,"given":"Mingze","family":"Gao","sequence":"additional","affiliation":[{"name":"College of Intelligence and Computing, Tianjin University, Tianjin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9714-0215","authenticated-orcid":false,"given":"Jiangtao","family":"Xie","sequence":"additional","affiliation":[{"name":"School of Information and Communication Engineering, Dalian University of Technology, Dalian, Liaoning, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4310-9140","authenticated-orcid":false,"given":"Pengfei","family":"Zhu","sequence":"additional","affiliation":[{"name":"College of Intelligence and Computing, Tianjin University, Tianjin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7229-3867","authenticated-orcid":false,"given":"Peihua","family":"Li","sequence":"additional","affiliation":[{"name":"School of Information and Communication Engineering, Dalian University of Technology, Dalian, Liaoning, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3330-783X","authenticated-orcid":false,"given":"Wangmeng","family":"Zuo","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7765-8095","authenticated-orcid":false,"given":"Qinghua","family":"Hu","sequence":"additional","affiliation":[{"name":"College of Intelligence and Computing, Tianjin University, Tianjin, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","first-page":"2488","article-title":"How does batch normalization help optimization?","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Santurkar"},{"key":"ref2","first-page":"7705","article-title":"Understanding batch normalization","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Bjorck"},{"key":"ref3","first-page":"4898","article-title":"Understanding the effective receptive field in deep convolutional neural networks","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Luo"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2021.3131813"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.339"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.170"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/d16-1044"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2017.689"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.168"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.228"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.5244\/C.31.117"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00105"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58555-6_1"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2020.2974833"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-020-01376-1"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1907369117"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1007\/11744047_45"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2014.2361137"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.325"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.63"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.41"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.743"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00335"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00474"},{"key":"ref25","first-page":"10347","article-title":"Training data-efficient image transformers & distillation through attention","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Touvron"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.1.1"},{"key":"ref27","article-title":"Finding flatter minima with SGD","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Jastrzebski"},{"key":"ref28","first-page":"6391","article-title":"Visualizing the loss landscape of neural nets","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Li"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00605"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.5555\/3045118.3045167"},{"key":"ref31","first-page":"2121","article-title":"Adaptive subgradient methods for online learning and stochastic optimization","volume":"12","author":"Duchi","year":"2011","journal-title":"J. Mach. Learn. Res."},{"key":"ref32","article-title":"Adam: A method for stochastic optimization","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Kingma"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01264-9_8"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.634"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00060"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00986"},{"key":"ref38","article-title":"DARTS: Differentiable architecture search","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Liu"},{"key":"ref39","first-page":"854","article-title":"Parseval networks: Improving robustness to adversarial examples","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Ciss\u00e9"},{"key":"ref40","article-title":"Boosting the certified robustness of L-infinity distance nets","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Zhang"},{"key":"ref41","article-title":"Explaining and harnessing adversarial examples","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Goodfellow"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.06083"},{"key":"ref43","article-title":"Benchmarking neural network robustness to common corruptions and perturbations","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Hendrycks"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.01501"},{"key":"ref45","article-title":"ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Geirhos"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.01078"},{"key":"ref47","first-page":"33576","article-title":"DropCov: A simple yet effective method for improving deep architectures","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Wang"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-013-0636-x"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1007\/11566465_15"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.480"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00115"},{"key":"ref52","article-title":"Fast differentiable matrix square root","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Song"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2022.3216339"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01216-8_38"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58529-7_13"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2022.3178802"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i04.5811"},{"key":"ref58","article-title":"A downsampled variant of ImageNet as an alternative to the CIFAR datasets","author":"Chrabaszcz","year":"2017"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2017.07.002"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1007\/s00365-021-09548-z"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-020-14663-9"},{"key":"ref62","first-page":"70:1","article-title":"The implicit bias of gradient descent on separable data","volume":"19","author":"Soudry","year":"2018","journal-title":"J. Mach. Learn. Res."},{"key":"ref63","first-page":"322","article-title":"Fine-grained analysis of optimization and generalization for overparameterized two-layer neural networks","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Arora"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.1145\/3446776"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10590-1_53"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2018.2858759"},{"key":"ref67","first-page":"2172","article-title":"InfoGAN: Interpretable representation learning by information maximizing generative adversarial nets","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Chen"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2020.3034267"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v36i2.20015"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.1631\/FITEE.1700808"},{"key":"ref71","first-page":"351","article-title":"Dropout training as adaptive regularization","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Wager"},{"key":"ref72","first-page":"2814","article-title":"Understanding dropout","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Baldi"},{"key":"ref73","article-title":"Very deep convolutional networks for large-scale image recognition","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Simonyan"},{"key":"ref74","article-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2009"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref76","article-title":"SoT: Delving deeper into classification head for transformer","author":"Xie","year":"2021"},{"key":"ref77","article-title":"Efficient and accurate estimation of Lipschitz constants for deep neural networks","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Fazlyab"},{"key":"ref78","first-page":"3839","article-title":"Lipschitz regularity of deep neural networks: Analysis and efficient estimation","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Virmaux"},{"key":"ref79","first-page":"291","article-title":"Sorting out Lipschitz function approximation","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Anil"},{"key":"ref80","article-title":"Training deep networks with structured layers by matrix backpropagation","author":"Ionescu","year":"2015"},{"key":"ref81","first-page":"918","article-title":"Second order properties of error surfaces","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"LeCun"},{"key":"ref82","article-title":"Rethinking the usage of batch normalization and Dropout in the training of deep neural networks","author":"Chen","year":"2019"},{"key":"ref83","article-title":"Reducing overfitting in deep networks by decorrelating representations","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Cogswell"},{"key":"ref84","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4615-7566-5"},{"issue":"1","key":"ref85","first-page":"1929","article-title":"Dropout: A simple way to prevent neural networks from overfitting","volume":"15","author":"Srivastava","year":"2014","journal-title":"J. Mach. Learn. Res."},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00745"},{"key":"ref87","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.01155"},{"key":"ref88","article-title":"Layer normalization","author":"Ba","year":"2016"},{"key":"ref89","doi-asserted-by":"publisher","DOI":"10.1088\/1742-5468\/ac9830"},{"key":"ref90","article-title":"ADADELTA: An adaptive learning rate method","author":"Zeiler","year":"2012"},{"key":"ref91","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v30i1.10200"},{"key":"ref92","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.01264"},{"key":"ref93","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-020-01396-x"},{"key":"ref94","first-page":"1106","article-title":"ImageNet classification with deep convolutional neural networks","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Krizhevsky"},{"key":"ref95","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.308"},{"key":"ref96","article-title":"An image is worth 16x16 words: Transformers for image recognition at scale","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Dosovitskiy"},{"key":"ref97","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v36i1.19967"},{"key":"ref98","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10602-1_48"},{"key":"ref99","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2016.2577031"},{"key":"ref100","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.322"},{"key":"ref101","article-title":"MMDetection: Open MMLab detection toolbox and benchmark","author":"Chen","year":"2019"},{"key":"ref102","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00914"},{"key":"ref103","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2019.2913372"}],"container-title":["IEEE Transactions on Pattern Analysis and Machine Intelligence"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/34\/10308548\/10269023.pdf?arnumber=10269023","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,3,2]],"date-time":"2024-03-02T07:32:55Z","timestamp":1709364775000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10269023\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,12]]},"references-count":103,"journal-issue":{"issue":"12"},"URL":"https:\/\/doi.org\/10.1109\/tpami.2023.3321392","relation":{},"ISSN":["0162-8828","2160-9292","1939-3539"],"issn-type":[{"value":"0162-8828","type":"print"},{"value":"2160-9292","type":"electronic"},{"value":"1939-3539","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,12]]}}}