{"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/inference-compilation-and-universal","title":"Inference Compilation and Universal Probabilistic Programming","arxiv_id":"1610.09900","date":"2016-10-31","proceeding":null,"authors":["Tuan Anh Le","Atilim Gunes Baydin","Frank Wood"],"abstract":"We introduce a method for using deep neural networks to amortize the cost of\ninference in models from the family induced by universal probabilistic\nprogramming languages, establishing a framework that combines the strengths of\nprobabilistic programming and deep learning methods. We call what we do\n\"compilation of inference\" because our method transforms a denotational\nspecification of an inference problem in the form of a probabilistic program\nwritten in a universal programming language into a trained neural network\ndenoted in a neural network specification language. When at test time this\nneural network is fed observational data and executed, it performs approximate\ninference in the original model specified by the probabilistic program. Our\ntraining objective and learning procedure are designed to allow the trained\nneural network to be used as a proposal distribution in a sequential importance\nsampling inference engine. We illustrate our method on mixture models and\nCaptcha solving and show significant speedups in the efficiency of inference.","url_abs":"http://arxiv.org/abs/1610.09900v2","url_pdf":"http://arxiv.org/pdf/1610.09900v2.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":"inference-compilation-and-universal","repo_url":"https://github.com/plai-group/pyprob","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"BSD-2-Clause"}},{"paper_slug":"inference-compilation-and-universal","repo_url":"https://github.com/probprog/pyprob","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"BSD-2-Clause"}},{"paper_slug":"inference-compilation-and-universal","repo_url":"https://github.com/pyprob/pyprob","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"BSD-2-Clause"}},{"paper_slug":"inference-compilation-and-universal","repo_url":"https://github.com/wsgharvey/pyro-infcomp-examples","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"probabilistic-programming","task_name":"Probabilistic Programming"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1610.09900","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}