{"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/implicit-reparameterization-gradients","title":"Implicit Reparameterization Gradients","arxiv_id":"1805.08498","date":"2018-05-22","proceeding":"NeurIPS 2018 12","authors":["Michael Figurnov","Shakir Mohamed","andriy mnih"],"abstract":"By providing a simple and efficient way of computing low-variance gradients\nof continuous random variables, the reparameterization trick has become the\ntechnique of choice for training a variety of latent variable models. However,\nit is not applicable to a number of important continuous distributions. We\nintroduce an alternative approach to computing reparameterization gradients\nbased on implicit differentiation and demonstrate its broader applicability by\napplying it to Gamma, Beta, Dirichlet, and von Mises distributions, which\ncannot be used with the classic reparameterization trick. Our experiments show\nthat the proposed approach is faster and more accurate than the existing\ngradient estimators for these distributions.","url_abs":"http://arxiv.org/abs/1805.08498v4","url_pdf":"http://arxiv.org/pdf/1805.08498v4.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":"implicit-reparameterization-gradients","repo_url":"https://github.com/lucadellalib/sac-beta","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"implicit-reparameterization-gradients","repo_url":"https://github.com/sophieburkhardt/dirichlet-vae-topic-models","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.08498","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}