{"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/meta-learning-for-stochastic-gradient-mcmc","title":"Meta-Learning for Stochastic Gradient MCMC","arxiv_id":"1806.04522","date":"2018-06-12","proceeding":"ICLR 2019 5","authors":["Wenbo Gong","Yingzhen Li","José Miguel Hernández-Lobato"],"abstract":"Stochastic gradient Markov chain Monte Carlo (SG-MCMC) has become\nincreasingly popular for simulating posterior samples in large-scale Bayesian\nmodeling. However, existing SG-MCMC schemes are not tailored to any specific\nprobabilistic model, even a simple modification of the underlying dynamical\nsystem requires significant physical intuition. This paper presents the first\nmeta-learning algorithm that allows automated design for the underlying\ncontinuous dynamics of an SG-MCMC sampler. The learned sampler generalizes\nHamiltonian dynamics with state-dependent drift and diffusion, enabling fast\ntraversal and efficient exploration of neural network energy landscapes.\nExperiments validate the proposed approach on both Bayesian fully connected\nneural network and Bayesian recurrent neural network tasks, showing that the\nlearned sampler out-performs generic, hand-designed SG-MCMC algorithms, and\ngeneralizes to different datasets and larger architectures.","url_abs":"http://arxiv.org/abs/1806.04522v1","url_pdf":"http://arxiv.org/pdf/1806.04522v1.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":"meta-learning-for-stochastic-gradient-mcmc","repo_url":"https://github.com/WenboGong/MetaSGMCMC","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"efficient-exploration","task_name":"Efficient Exploration"},{"task_slug":"meta-learning","task_name":"Meta-Learning"},{"task_slug":"physical-intuition","task_name":"Physical Intuition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1806.04522","atlas_url":"https://app.syntology.ai/?focus=1806.04522","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}