{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/dp-em-differentially-private-expectation","title":"DP-EM: Differentially Private Expectation Maximization","arxiv_id":"1605.06995","date":"2016-05-23","proceeding":null,"authors":["Mijung Park","Jimmy Foulds","Kamalika Chaudhuri","Max Welling"],"abstract":"The iterative nature of the expectation maximization (EM) algorithm presents\na challenge for privacy-preserving estimation, as each iteration increases the\namount of noise needed. We propose a practical private EM algorithm that\novercomes this challenge using two innovations: (1) a novel moment perturbation\nformulation for differentially private EM (DP-EM), and (2) the use of two\nrecently developed composition methods to bound the privacy \"cost\" of multiple\nEM iterations: the moments accountant (MA) and zero-mean concentrated\ndifferential privacy (zCDP). Both MA and zCDP bound the moment generating\nfunction of the privacy loss random variable and achieve a refined tail bound,\nwhich effectively decrease the amount of additive noise. We present empirical\nresults showing the benefits of our approach, as well as similar performance\nbetween these two composition methods in the DP-EM setting for Gaussian mixture\nmodels. Our approach can be readily extended to many iterative learning\nalgorithms, opening up various exciting future directions.","url_abs":"http://arxiv.org/abs/1605.06995v2","url_pdf":"http://arxiv.org/pdf/1605.06995v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"dp-em-differentially-private-expectation","repo_url":"https://github.com/thehimalayanleo/Private-Generative-Models","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"privacy-preserving","task_name":"Privacy Preserving"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}