{"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/variational-bayes-in-private-settings-vips","title":"Variational Bayes In Private Settings (VIPS)","arxiv_id":"1611.00340","date":"2016-11-01","proceeding":null,"authors":["Mijung Park","James Foulds","Kamalika Chaudhuri","Max Welling"],"abstract":"Many applications of Bayesian data analysis involve sensitive information,\nmotivating methods which ensure that privacy is protected. We introduce a\ngeneral privacy-preserving framework for Variational Bayes (VB), a widely used\noptimization-based Bayesian inference method. Our framework respects\ndifferential privacy, the gold-standard privacy criterion, and encompasses a\nlarge class of probabilistic models, called the Conjugate Exponential (CE)\nfamily. We observe that we can straightforwardly privatise VB's approximate\nposterior distributions for models in the CE family, by perturbing the expected\nsufficient statistics of the complete-data likelihood. For a broadly-used class\nof non-CE models, those with binomial likelihoods, we show how to bring such\nmodels into the CE family, such that inferences in the modified model resemble\nthe private variational Bayes algorithm as closely as possible, using the\nPolya-Gamma data augmentation scheme. The iterative nature of variational Bayes\npresents a further challenge since iterations increase the amount of noise\nneeded. We overcome this by combining: (1) an improved composition method for\ndifferential privacy, called the moments accountant, which provides a tight\nbound on the privacy cost of multiple VB iterations and thus significantly\ndecreases the amount of additive noise; and (2) the privacy amplification\neffect of subsampling mini-batches from large-scale data in stochastic\nlearning. We empirically demonstrate the effectiveness of our method in CE and\nnon-CE models including latent Dirichlet allocation, Bayesian logistic\nregression, and sigmoid belief networks, evaluated on real-world datasets.","url_abs":"http://arxiv.org/abs/1611.00340v5","url_pdf":"http://arxiv.org/pdf/1611.00340v5.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":"variational-bayes-in-private-settings-vips","repo_url":"https://github.com/mijungi/vips_code","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"bayesian-inference","task_name":"Bayesian Inference"},{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"privacy-preserving","task_name":"Privacy Preserving"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.00340","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}