{"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/linear-response-methods-for-accurate","title":"Linear Response Methods for Accurate Covariance Estimates from Mean Field Variational Bayes","arxiv_id":"1506.04088","date":"2015-06-12","proceeding":"NeurIPS 2015 12","authors":["Ryan Giordano","Tamara Broderick","Michael Jordan"],"abstract":"Mean field variational Bayes (MFVB) is a popular posterior approximation\nmethod due to its fast runtime on large-scale data sets. However, it is well\nknown that a major failing of MFVB is that it underestimates the uncertainty of\nmodel variables (sometimes severely) and provides no information about model\nvariable covariance.\n  We generalize linear response methods from statistical physics to deliver\naccurate uncertainty estimates for model variables---both for individual\nvariables and coherently across variables. We call our method linear response\nvariational Bayes (LRVB). When the MFVB posterior approximation is in the\nexponential family, LRVB has a simple, analytic form, even for non-conjugate\nmodels. Indeed, we make no assumptions about the form of the true posterior. We\ndemonstrate the accuracy and scalability of our method on a range of models for\nboth simulated and real data.","url_abs":"http://arxiv.org/abs/1506.04088v2","url_pdf":"http://arxiv.org/pdf/1506.04088v2.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":"linear-response-methods-for-accurate","repo_url":"https://github.com/rgiordan/LinearResponseVariationalBayesNIPS2015","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"GPL-2.0"}}],"tasks":[{"task_slug":"form","task_name":"Form"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1506.04088","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}