{"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/a-nice-mc-adversarial-training-for-mcmc","title":"A-NICE-MC: Adversarial Training for MCMC","arxiv_id":"1706.07561","date":"2017-06-23","proceeding":"NeurIPS 2017 12","authors":["Jiaming Song","Shengjia Zhao","Stefano Ermon"],"abstract":"Existing Markov Chain Monte Carlo (MCMC) methods are either based on\ngeneral-purpose and domain-agnostic schemes which can lead to slow convergence,\nor hand-crafting of problem-specific proposals by an expert. We propose\nA-NICE-MC, a novel method to train flexible parametric Markov chain kernels to\nproduce samples with desired properties. First, we propose an efficient\nlikelihood-free adversarial training method to train a Markov chain and mimic a\ngiven data distribution. Then, we leverage flexible volume preserving flows to\nobtain parametric kernels for MCMC. Using a bootstrap approach, we show how to\ntrain efficient Markov chains to sample from a prescribed posterior\ndistribution by iteratively improving the quality of both the model and the\nsamples. A-NICE-MC provides the first framework to automatically design\nefficient domain-specific MCMC proposals. Empirical results demonstrate that\nA-NICE-MC combines the strong guarantees of MCMC with the expressiveness of\ndeep neural networks, and is able to significantly outperform competing methods\nsuch as Hamiltonian Monte Carlo.","url_abs":"http://arxiv.org/abs/1706.07561v3","url_pdf":"http://arxiv.org/pdf/1706.07561v3.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":"a-nice-mc-adversarial-training-for-mcmc","repo_url":"https://github.com/ermongroup/a-nice-mc","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"a-nice-mc-adversarial-training-for-mcmc","repo_url":"https://github.com/jiamings/a-nice-mc","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"a-nice-mc-adversarial-training-for-mcmc","repo_url":"https://github.com/GallupGovt/multivac","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.07561","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}