{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T15:13:30Z","timestamp":1784646810333,"version":"3.55.0"},"reference-count":60,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2025,1,1]],"date-time":"2025-01-01T00:00:00Z","timestamp":1735689600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2025,1,1]],"date-time":"2025-01-01T00:00:00Z","timestamp":1735689600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,1,1]],"date-time":"2025-01-01T00:00:00Z","timestamp":1735689600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62302441"],"award-info":[{"award-number":["62302441"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Key Research and Development Plans of Guangxi Province","award":["AB22080077"],"award-info":[{"award-number":["AB22080077"]}]},{"name":"Information Technology Center of Zhejiang University and the Supercomputing Center of Hangzhou City University"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans.Inform.Forensic Secur."],"published-print":{"date-parts":[[2025]]},"DOI":"10.1109\/tifs.2025.3557755","type":"journal-article","created":{"date-parts":[[2025,4,18]],"date-time":"2025-04-18T17:36:19Z","timestamp":1744997779000},"page":"4498-4511","source":"Crossref","is-referenced-by-count":11,"title":["DC-SGD: Differentially Private SGD With Dynamic Clipping Through Gradient Norm Distribution Estimation"],"prefix":"10.1109","volume":"20","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8849-8808","authenticated-orcid":false,"given":"Chengkun","family":"Wei","sequence":"first","affiliation":[{"name":"College of Computer Science and Technology, Zhejiang University, Hangzhou, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Weixian","family":"Li","sequence":"additional","affiliation":[{"name":"Ant Group, Hangzhou, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chen","family":"Gong","sequence":"additional","affiliation":[{"name":"School of Engineering, University of Virginia, Charlottesville, VA, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1674-4701","authenticated-orcid":false,"given":"Wenzhi","family":"Chen","sequence":"additional","affiliation":[{"name":"College of Computer Science and Technology, Zhejiang University, Hangzhou, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00584"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1016\/j.patrec.2020.07.042"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i07.7003"},{"key":"ref4","first-page":"2249","article-title":"Cascaded diffusion models for high fidelity image generation","volume":"23","author":"Ho","year":"2022","journal-title":"J. Mach. Learn. Res."},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/SP54263.2024.00224"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2020.2979670"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-020-18073-9"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1016\/j.jksuci.2020.04.001"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2020.2981314"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/SP.2017.41"},{"key":"ref11","first-page":"1964","article-title":"Label-only membership inference attacks","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Choquette-Choo"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.2478\/popets-2019-0008"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/2810103.2813677"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1145\/3564625.3564636"},{"key":"ref15","article-title":"PrivImage: Differentially private synthetic image generation using diffusion models with semantic-aware pretraining","author":"Li","year":"2023","journal-title":"arXiv:2311.12850"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1016\/j.compbiolchem.2020.107310"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.56553\/popets-2025-0068"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/2976749.2978318"},{"key":"ref19","first-page":"17455","article-title":"Differentially private learning with adaptive clipping","volume-title":"Proc. Annu. Conf. Neural Inf. Process. Syst.","volume":"34","author":"Andrew"},{"key":"ref20","article-title":"Unlocking high-accuracy differentially private image classification through scale","author":"De","year":"2022","journal-title":"arXiv:2204.13650"},{"key":"ref21","article-title":"Automatic clipping: Differentially private deep learning made easier and stronger","author":"Bu","year":"2022","journal-title":"arXiv:2206.07136"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.00819"},{"key":"ref23","article-title":"Dynamic differential-privacy preserving SGD","author":"Du","year":"2021","journal-title":"arXiv:2111.00173"},{"key":"ref24","first-page":"13773","article-title":"Understanding gradient clipping in private SGD: A geometric perspective","volume-title":"Proc. Annu. Conf. Neural Inf. Process. Syst.","volume":"33","author":"Chen"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3313276.3316377"},{"key":"ref26","article-title":"Hyperparameter tuning with Renyi differential privacy","author":"Papernot","year":"2021","journal-title":"arXiv:2110.03620"},{"key":"ref27","article-title":"AdaCliP: Adaptive clipping for private SGD","author":"Pichapati","year":"2019","journal-title":"arXiv:1908.07643"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1111\/rssb.12454"},{"key":"ref29","article-title":"Adam: A method for stochastic optimization","author":"Kingma","year":"2014","journal-title":"arXiv:1412.6980"},{"key":"ref30","volume-title":"MNIST Handwritten Digit Database","author":"LeCun","year":"2010"},{"key":"ref31","article-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2009"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.2118\/18761-MS"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W18-5446"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref35","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":"2018","journal-title":"arXiv:1810.04805"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/CSF.2017.11"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/SP.2019.00019"},{"key":"ref38","article-title":"Differentially private learning needs better features (or much more data)","author":"Tram\u00e8r","year":"2020","journal-title":"arXiv:2011.11660"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1007\/11681878_14"},{"key":"ref40","article-title":"Learning differentially private recurrent language models","author":"Brendan McMahan","year":"2017","journal-title":"arXiv:1710.06963"},{"key":"ref41","first-page":"2496","article-title":"Hypothesis testing interpretations and Renyi differential privacy","volume-title":"Proc. Int. Conf. Artif. Intell. Statist. (AISTATS)","author":"Balle"},{"key":"ref42","article-title":"R\u00e9nyi differential privacy of the sampled Gaussian mechanism","author":"Mironov","year":"2019","journal-title":"arXiv:1908.10530"},{"key":"ref43","article-title":"Differentially private deep learning with ModelMix","author":"Xiao","year":"2022","journal-title":"arXiv:2210.03843"},{"key":"ref44","article-title":"Do not let privacy overbill utility: Gradient embedding perturbation for private learning","author":"Yu","year":"2021","journal-title":"arXiv:2102.12677"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1145\/3548606.3560562"},{"key":"ref46","article-title":"Evading curse of dimensionality in unconstrained private GLMs via private gradient descent","author":"Song","year":"2020","journal-title":"arXiv:2006.06783"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.29012\/jpc.870"},{"key":"ref48","first-page":"38","article-title":"Transformers: State-of-the-art natural language processing","volume-title":"Proc. Conf. Empirical Methods Natural Lang. Syst. Demonstration","author":"Wolf"},{"key":"ref49","first-page":"1","article-title":"PyTorch: An imperative style, high-performance deep learning library","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Paszke"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v36i7.20749"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.29012\/jpc.880"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v35i10.17123"},{"key":"ref53","first-page":"12208","article-title":"Large scale private learning via low-rank reparametrization","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Yu"},{"key":"ref54","article-title":"DPDR: Gradient decomposition and reconstruction for differentially private deep learning","author":"Liu","year":"2024","journal-title":"arXiv:2406.02744"},{"key":"ref55","article-title":"Clip body and tail separately: High probability guarantees for DPSGD with heavy tails","author":"Sha","year":"2024","journal-title":"arXiv:2405.17529"},{"key":"ref56","article-title":"Normalized\/Clipped SGD with perturbation for differentially private non-convex optimization","author":"Yang","year":"2022","journal-title":"arXiv:2206.13033"},{"key":"ref57","first-page":"263","article-title":"Bounding user contributions: A bias-variance trade-off in differential privacy","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Amin"},{"key":"ref58","first-page":"1","article-title":"Understanding clipping for federated learning: Convergence and client-level differential privacy","volume-title":"Proc. Int. Conf. Mach. Learn., (ICML)","author":"Zhang"},{"key":"ref59","article-title":"Differentially private zeroth-order methods for scalable large language model finetuning","author":"Liu","year":"2024","journal-title":"arXiv:2402.07818"},{"key":"ref60","first-page":"1","article-title":"DPZero: Private fine-tuning of language models without backpropagation","volume-title":"Proc. 41st Int. Conf. Mach. Learn.","author":"Zhang"}],"container-title":["IEEE Transactions on Information Forensics and Security"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10206\/10810755\/10969624.pdf?arnumber=10969624","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,5,7]],"date-time":"2025-05-07T17:52:33Z","timestamp":1746640353000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10969624\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025]]},"references-count":60,"URL":"https:\/\/doi.org\/10.1109\/tifs.2025.3557755","relation":{},"ISSN":["1556-6013","1556-6021"],"issn-type":[{"value":"1556-6013","type":"print"},{"value":"1556-6021","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025]]}}}