{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T12:10:50Z","timestamp":1780661450162,"version":"3.54.1"},"reference-count":45,"publisher":"Wiley","issue":"6","license":[{"start":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T00:00:00Z","timestamp":1765411200000},"content-version":"vor","delay-in-days":10,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Stat Anal Data Min: An ASA Data Sci Journal"],"published-print":{"date-parts":[[2025,12]]},"abstract":"<jats:title>ABSTRACT<\/jats:title>\n                  <jats:p>Recent advances have sparked significant interest in the development of privacy\u2010preserving Principal Component Analysis (PCA). However, many existing approaches rely on restrictive assumptions, such as assuming sub\u2010Gaussian data or being vulnerable to data contamination. Additionally, some methods are computationally expensive or depend on unknown model parameters that must be estimated, limiting their accessibility for data analysts seeking privacy\u2010preserving PCA. In this paper, we propose a differentially private PCA method applicable to heavy\u2010tailed and potentially contaminated data. Our approach leverages the property that the covariance matrix of properly rescaled data preserves eigenvectors and their order under elliptical distributions, which include Gaussian and heavy\u2010tailed distributions. By applying a bounded transformation, we enable straightforward computation of principal components in a differentially private manner. Additionally, boundedness guarantees robustness against data contamination. We conduct both theoretical analysis and empirical evaluations of the proposed method, focusing on its ability to recover the subspace spanned by the leading principal components. Extensive numerical experiments demonstrate that our method consistently outperforms existing approaches in terms of statistical utility, particularly in non\u2010Gaussian or contaminated data settings.<\/jats:p>","DOI":"10.1002\/sam.70053","type":"journal-article","created":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T12:21:32Z","timestamp":1765455692000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Robust and Differentially Private Principal Component Analysis"],"prefix":"10.1002","volume":"18","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-3220-6504","authenticated-orcid":false,"given":"Minwoo","family":"Kim","sequence":"first","affiliation":[{"name":"Department of Statistics Seoul National University  Seoul South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6023-8956","authenticated-orcid":false,"given":"Sungkyu","family":"Jung","sequence":"additional","affiliation":[{"name":"Department of Statistics Seoul National University  Seoul South Korea"},{"name":"Institute for Data Innovation in Science Seoul National University  Seoul South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2025,12,11]]},"reference":[{"key":"e_1_2_9_2_1","doi-asserted-by":"crossref","unstructured":"R.Cummings D.Desfontaines D.Evans et al. \u201cAdvancing Differential Privacy: Where We Are Now and Future Directions for Real\u2010World Deployment \u201d2024Harvard Data Science Review 6 https:\/\/hdsr.mitpress.mit.edu\/pub\/sl9we8gh.","DOI":"10.1162\/99608f92.d3197524"},{"key":"e_1_2_9_3_1","doi-asserted-by":"publisher","DOI":"10.1146\/annurev-statistics-112723-034158"},{"key":"e_1_2_9_4_1","doi-asserted-by":"publisher","DOI":"10.1007\/11681878_14"},{"key":"e_1_2_9_5_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-662-53641-4_24"},{"key":"e_1_2_9_6_1","doi-asserted-by":"publisher","DOI":"10.1111\/rssb.12454"},{"key":"e_1_2_9_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/CSF.2017.11"},{"key":"e_1_2_9_8_1","doi-asserted-by":"publisher","DOI":"10.1561\/0400000042"},{"key":"e_1_2_9_9_1","doi-asserted-by":"publisher","DOI":"10.1145\/1536414.1536466"},{"key":"e_1_2_9_10_1","doi-asserted-by":"publisher","DOI":"10.1109\/FOCS.2007.66"},{"key":"e_1_2_9_11_1","doi-asserted-by":"publisher","DOI":"10.1145\/1250790.1250803"},{"key":"e_1_2_9_12_1","doi-asserted-by":"publisher","DOI":"10.1145\/2976749.2978318"},{"key":"e_1_2_9_13_1","doi-asserted-by":"publisher","DOI":"10.1214\/25-EJS2349"},{"key":"e_1_2_9_14_1","doi-asserted-by":"publisher","DOI":"10.1214\/25-EJS2368"},{"key":"e_1_2_9_15_1","doi-asserted-by":"publisher","DOI":"10.1002\/sta4.658"},{"key":"e_1_2_9_16_1","unstructured":"Y.\u2010X.Wang \u201cRevisiting Differentially Private Linear Regression: Optimal and Adaptive Prediction & Estimation in Unbounded Domain \u201d2018arXiv preprint arXiv: 1803.02596."},{"key":"e_1_2_9_17_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-11257-2_14"},{"key":"e_1_2_9_18_1","unstructured":"X.Chen C.Liu B.Li K.Lu andD.Song \u201cTargeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning \u201d2017arXiv:1712.05526."},{"key":"e_1_2_9_19_1","first-page":"1689","volume-title":"Proceedings of the 32nd International Conference on Machine Learning","author":"Xiao H.","year":"2015"},{"key":"e_1_2_9_20_1","first-page":"3887","volume-title":"Advances in Neural Information Processing Systems","author":"Liu X.","year":"2021"},{"key":"e_1_2_9_21_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jmva.2018.11.010"},{"key":"e_1_2_9_22_1","doi-asserted-by":"publisher","DOI":"10.1080\/01621459.2016.1246366"},{"key":"e_1_2_9_23_1","first-page":"173","volume-title":"Proceedings of Mathematical and Scientific Machine Learning","author":"Maunu T.","year":"2022"},{"key":"e_1_2_9_24_1","first-page":"1167","volume-title":"Proceedings of Thirty Fifth Conference on Learning Theory","author":"Liu X.","year":"2022"},{"key":"e_1_2_9_25_1","first-page":"1121","volume-title":"Proceedings of the 40th International Conference on Machine Learning","author":"Asi H.","year":"2023"},{"key":"e_1_2_9_26_1","first-page":"14106","volume-title":"Advances in Neural Information Processing Systems","author":"Asi H.","year":"2020"},{"key":"e_1_2_9_27_1","doi-asserted-by":"publisher","DOI":"10.1111\/insr.12113"},{"key":"e_1_2_9_28_1","doi-asserted-by":"publisher","DOI":"10.1016\/S0378-3758(00)00199-3"},{"key":"e_1_2_9_29_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11222-024-10413-9"},{"key":"e_1_2_9_30_1","doi-asserted-by":"publisher","DOI":"10.1201\/9781351077040"},{"key":"e_1_2_9_31_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-0348-8201-9_22"},{"key":"e_1_2_9_32_1","doi-asserted-by":"publisher","DOI":"10.1016\/S0167-7152(98)00272-7"},{"key":"e_1_2_9_33_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.spl.2012.01.001"},{"key":"e_1_2_9_34_1","doi-asserted-by":"publisher","DOI":"10.1214\/17-AOS1541"},{"key":"e_1_2_9_35_1","first-page":"3061","volume-title":"Proceedings of the 28th International Conference on Artificial Intelligence and Statistics","author":"Han S.","year":"2025"},{"key":"e_1_2_9_36_1","doi-asserted-by":"publisher","DOI":"10.1093\/biomet\/asv008"},{"key":"e_1_2_9_37_1","volume-title":"Robust Statistics: The Approach Based on Influence Functions","author":"Hampel F.","year":"1986"},{"key":"e_1_2_9_38_1","volume-title":"Robust Statistics","author":"Huber P. J.","year":"2011"},{"key":"e_1_2_9_39_1","doi-asserted-by":"publisher","DOI":"10.1145\/2591796.2591883"},{"key":"e_1_2_9_40_1","first-page":"29929","volume-title":"Advances in Neural Information Processing Systems","author":"Liu X.","year":"2022"},{"key":"e_1_2_9_41_1","unstructured":"T. T.Cai D.Xia andM.Zha \u201cOptimal Differentially Private PCA and Estimation for Spiked Covariance Matrices \u201d2024arXiv:2401.03820."},{"key":"e_1_2_9_42_1","doi-asserted-by":"publisher","DOI":"10.1038\/nature07331"},{"key":"e_1_2_9_43_1","first-page":"14475","volume-title":"Advances in Neural Information Processing Systems","author":"Biswas S.","year":"2020"},{"key":"e_1_2_9_44_1","doi-asserted-by":"publisher","DOI":"10.1017\/9781108231596"},{"key":"e_1_2_9_45_1","unstructured":"K.RamsayandD.Spicker \u201cImproved Subsample\u2010and\u2010Aggregate via the Private Modified Winsorized Mean \u201d2025arXiv preprint arXiv: 2501.14095."},{"key":"e_1_2_9_46_1","first-page":"77691","volume-title":"Advances in Neural Information Processing Systems","author":"Durfee D.","year":"2023"}],"container-title":["Statistical Analysis and Data Mining: An ASA Data Science Journal"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/sam.70053","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T04:29:50Z","timestamp":1768192190000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1002\/sam.70053"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12]]},"references-count":45,"journal-issue":{"issue":"6","published-print":{"date-parts":[[2025,12]]}},"alternative-id":["10.1002\/sam.70053"],"URL":"https:\/\/doi.org\/10.1002\/sam.70053","archive":["Portico"],"relation":{},"ISSN":["1932-1864","1932-1872"],"issn-type":[{"value":"1932-1864","type":"print"},{"value":"1932-1872","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12]]},"assertion":[{"value":"2025-05-30","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-11-24","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-12-11","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70053"}}