{"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/conjugate-computation-variational-inference","title":"Conjugate-Computation Variational Inference : Converting Variational Inference in Non-Conjugate Models to Inferences in Conjugate Models","arxiv_id":"1703.04265","date":"2017-03-13","proceeding":null,"authors":["Mohammad Emtiyaz Khan","Wu Lin"],"abstract":"Variational inference is computationally challenging in models that contain\nboth conjugate and non-conjugate terms. Methods specifically designed for\nconjugate models, even though computationally efficient, find it difficult to\ndeal with non-conjugate terms. On the other hand, stochastic-gradient methods\ncan handle the non-conjugate terms but they usually ignore the conjugate\nstructure of the model which might result in slow convergence. In this paper,\nwe propose a new algorithm called Conjugate-computation Variational Inference\n(CVI) which brings the best of the two worlds together -- it uses conjugate\ncomputations for the conjugate terms and employs stochastic gradients for the\nrest. We derive this algorithm by using a stochastic mirror-descent method in\nthe mean-parameter space, and then expressing each gradient step as a\nvariational inference in a conjugate model. We demonstrate our algorithm's\napplicability to a large class of models and establish its convergence. Our\nexperimental results show that our method converges much faster than the\nmethods that ignore the conjugate structure of the model.","url_abs":"http://arxiv.org/abs/1703.04265v2","url_pdf":"http://arxiv.org/pdf/1703.04265v2.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":"conjugate-computation-variational-inference","repo_url":"https://github.com/emtiyaz/cvi","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"conjugate-computation-variational-inference","repo_url":"https://github.com/vmasrani/CVI_PLDS","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.04265","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}