{"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/lost-relatives-of-the-gumbel-trick","title":"Lost Relatives of the Gumbel Trick","arxiv_id":"1706.04161","date":"2017-06-13","proceeding":"ICML 2017 8","authors":["Matej Balog","Nilesh Tripuraneni","Zoubin Ghahramani","Adrian Weller"],"abstract":"The Gumbel trick is a method to sample from a discrete probability\ndistribution, or to estimate its normalizing partition function. The method\nrelies on repeatedly applying a random perturbation to the distribution in a\nparticular way, each time solving for the most likely configuration. We derive\nan entire family of related methods, of which the Gumbel trick is one member,\nand show that the new methods have superior properties in several settings with\nminimal additional computational cost. In particular, for the Gumbel trick to\nyield computational benefits for discrete graphical models, Gumbel\nperturbations on all configurations are typically replaced with so-called\nlow-rank perturbations. We show how a subfamily of our new methods adapts to\nthis setting, proving new upper and lower bounds on the log partition function\nand deriving a family of sequential samplers for the Gibbs distribution.\nFinally, we balance the discussion by showing how the simpler analytical form\nof the Gumbel trick enables additional theoretical results.","url_abs":"http://arxiv.org/abs/1706.04161v1","url_pdf":"http://arxiv.org/pdf/1706.04161v1.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":"lost-relatives-of-the-gumbel-trick","repo_url":"https://github.com/matejbalog/gumbel-relatives","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1706.04161","atlas_url":"https://app.syntology.ai/?focus=1706.04161","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}