{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T15:31:04Z","timestamp":1772119864478,"version":"3.50.1"},"reference-count":42,"publisher":"Springer Science and Business Media LLC","issue":"3","license":[{"start":{"date-parts":[[2023,4,4]],"date-time":"2023-04-04T00:00:00Z","timestamp":1680566400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,4,4]],"date-time":"2023-04-04T00:00:00Z","timestamp":1680566400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Stat Comput"],"published-print":{"date-parts":[[2023,6]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Stochastic gradient Markov chain Monte Carlo (SGMCMC) is a popular class of algorithms for scalable Bayesian inference. However, these algorithms include hyperparameters such as step size or batch size that influence the accuracy of estimators based on the obtained posterior samples. As a result, these hyperparameters must be tuned by the practitioner and currently no principled and automated way to tune them exists. Standard Markov chain Monte Carlo tuning methods based on acceptance rates cannot be used for SGMCMC, thus requiring alternative tools and diagnostics. We propose a novel bandit-based algorithm that tunes the SGMCMC hyperparameters by minimizing the Stein discrepancy between the true posterior and its Monte Carlo approximation. We provide theoretical results supporting this approach and assess various Stein-based discrepancies. We support our results with experiments on both simulated and real datasets, and find that this method is practical for a wide range of applications.<\/jats:p>","DOI":"10.1007\/s11222-023-10233-3","type":"journal-article","created":{"date-parts":[[2023,4,5]],"date-time":"2023-04-05T21:58:54Z","timestamp":1680731934000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Efficient and generalizable tuning strategies for stochastic gradient MCMC"],"prefix":"10.1007","volume":"33","author":[{"given":"Jeremie","family":"Coullon","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Leah","family":"South","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Christopher","family":"Nemeth","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"297","published-online":{"date-parts":[[2023,4,4]]},"reference":[{"issue":"4","key":"10233_CR1","doi-asserted-by":"publisher","first-page":"343","DOI":"10.1007\/s11222-008-9110-y","volume":"18","author":"C Andrieu","year":"2008","unstructured":"Andrieu, C., Thoms, J.: A tutorial on adaptive MCMC. Stat. Comput. 18(4), 343\u2013373 (2008)","journal-title":"Stat. Comput."},{"key":"10233_CR2","unstructured":"Audibert, J.Y., Bubeck, S., Munos, R.: Best arm identification in multi-armed bandits. In: COLT, pp. 41\u201353 (2010)"},{"issue":"3","key":"10233_CR3","doi-asserted-by":"publisher","first-page":"599","DOI":"10.1007\/s11222-018-9826-2","volume":"29","author":"J Baker","year":"2019","unstructured":"Baker, J., Fearnhead, P., Fox, E.B., et al.: Control variates for stochastic gradient MCMC. Stat. Comput. 29(3), 599\u2013615 (2019)","journal-title":"Stat. Comput."},{"key":"10233_CR4","unstructured":"Bingham, E., Chen, J.P., Jankowiak, M.: et\u00a0al. Pyro: Deep Universal Probabilistic Programming. arXiv preprint arXiv:1810.09538 (2018)"},{"key":"10233_CR5","unstructured":"Brosse, N., Durmus, A., Moulines, \u00c9.: The promises and pitfalls of stochastic gradient Langevin dynamics. In: Advances in Neural Information Processing Systems, pp. 8278\u20138288 (2018)"},{"key":"10233_CR6","doi-asserted-by":"crossref","unstructured":"Bubeck, S., Cesa-Bianchi, N.: Regret analysis of stochastic and nonstochastic multi-armed bandit problems (2012) arXiv preprint arXiv:1204.5721","DOI":"10.1561\/9781601986276"},{"key":"10233_CR7","unstructured":"Chen, C., Carlson, D., Gan, Z., et\u00a0al. Bridging the gap between stochastic gradient MCMC and stochastic optimization. In: Artificial Intelligence and Statistics, pp. 1051\u20131060 (2016)"},{"key":"10233_CR8","unstructured":"Chen, T., Fox, E., Guestrin, C.: Stochastic gradient Hamiltonian Monte Carlo. In: International Conference on Machine Learning, pp. 1683\u20131691 (2014)"},{"key":"10233_CR9","unstructured":"Chwialkowski, K., Strathmann, H., Gretton, A.: A kernel test of goodness of fit. In: International Conference on Machine Learning, pp. 2606\u20132615 (2016)"},{"issue":"72","key":"10233_CR10","doi-asserted-by":"publisher","first-page":"4113","DOI":"10.21105\/joss.04113","volume":"7","author":"J Coullon","year":"2022","unstructured":"Coullon, J., Nemeth, C.: SGMCMCJax: a lightweight JAX library for stochastic gradient Markov chain Monte Carlo algorithms. J. Open Source Softw. 7(72), 4113 (2022)","journal-title":"J. Open Source Softw."},{"key":"10233_CR11","unstructured":"Ding, N., Fang, Y., Babbush, R.: et\u00a0al Bayesian sampling using stochastic gradient thermostats. In: Advances in Neural Information Processing Systems, pp. 3203\u20133211 (2014a)"},{"key":"10233_CR12","unstructured":"Ding, N., Fang, Y., Babbush, R., et\u00a0al Bayesian sampling using stochastic gradient thermostats. In: Ghahramani Z, Welling M, Cortes C, et\u00a0al (eds.) Advances in Neural Information Processing Systems, vol.\u00a027. Curran Associates, Inc (2014b) https:\/\/proceedings.neurips.cc\/paper\/2014\/file\/21fe5b8ba755eeaece7a450849876228-Paper.pdf"},{"issue":"1","key":"10233_CR13","doi-asserted-by":"publisher","first-page":"110","DOI":"10.1214\/aoap\/1034625254","volume":"7","author":"A Gelman","year":"1997","unstructured":"Gelman, A., Gilks, W.R., Roberts, G.O.: Weak convergence and optimal scaling of random walk metropolis algorithms. Ann. Appl. Probab. 7(1), 110\u2013120 (1997)","journal-title":"Ann. Appl. Probab."},{"key":"10233_CR14","unstructured":"Gong, W., Li, Y., Hern\u00e1ndez-Lobato, J.M.: Sliced kernelized Stein discrepancy. CoRR abs\/2006.16531. (2020) https:\/\/arxiv.org\/abs\/2006.16531"},{"key":"10233_CR15","unstructured":"Gorham, J., Mackey, L.: Measuring sample quality with Stein\u2019s method. In: Advances in Neural Information Processing Systems, pp. 226\u2013234 (2015)"},{"key":"10233_CR16","unstructured":"Gorham, J., Mackey, L.: Measuring sample quality with kernels. In: Proceedings of the 34th International Conference on Machine Learning-Volume 70, JMLR. org, pp. 1292\u20131301 (2017)"},{"key":"10233_CR17","unstructured":"Gorham, J., Raj, A., Mackey, L.: Stochastic Stein discrepancies. (2020) arXiv preprint arXiv:2007.02857"},{"key":"10233_CR18","unstructured":"Guo, C., Pleiss, G., Sun, Y., et\u00a0al. On calibration of modern neural networks. In: Precup, D., Teh, Y.W. (eds) Proceedings of the 34th International Conference on Machine Learning, Proceedings of Machine Learning Research, vol\u00a070. pp. 1321\u20131330 (2017) http:\/\/proceedings.mlr.press\/v70\/guo17a.html"},{"key":"10233_CR19","doi-asserted-by":"publisher","unstructured":"Harper, F.M., Konstan, J.A.: The movielens datasets: history and context 5(4). (2015) https:\/\/doi.org\/10.1145\/2827872","DOI":"10.1145\/2827872"},{"issue":"1","key":"10233_CR20","first-page":"1593","volume":"15","author":"MD Hoffman","year":"2014","unstructured":"Hoffman, M.D., Gelman, A.: The no-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. J. Mach. Learn. Res. 15(1), 1593\u20131623 (2014)","journal-title":"J. Mach. Learn. Res."},{"key":"10233_CR21","unstructured":"Izmailov, P., Vikram, S., Hoffman, M.D., et\u00a0al.: What are bayesian neural network posteriors really like? (2021) arXiv preprint arXiv:2104.14421"},{"key":"10233_CR22","unstructured":"Jamieson, K., Talwalkar, A.: Non-stochastic best arm identification and hyperparameter optimization. In: Artificial Intelligence and Statistics, pp. 240\u2013248 (2016)"},{"key":"10233_CR23","unstructured":"Jitkrittum, W., Xu, W., Szab\u00f3, Z., et\u00a0al. A linear-time kernel goodness-of-fit test. (2017) arXiv preprint arXiv:1705.07673"},{"key":"10233_CR24","unstructured":"Karnin, Z., Koren, T., Somekh, O.: Almost optimal exploration in multi-armed bandits. In: International Conference on Machine Learning, pp. 1238\u20131246 (2013)"},{"key":"10233_CR25","unstructured":"Kim, S., Song, Q., Liang, F.: Stochastic gradient langevin dynamics algorithms with adaptive drifts. (2020) arXiv preprint arXiv:2009.09535"},{"key":"10233_CR26","unstructured":"Lakshminarayanan, B., Pritzel, A., Blundell, C.: Simple and scalable predictive uncertainty estimation using deep ensembles. In: Guyon, I., Luxburg, U.V., Bengio, S., et\u00a0al (eds) Advances in Neural Information Processing Systems, vol\u00a030. Curran Associates, Inc., (2017) https:\/\/proceedings.neurips.cc\/paper\/2017\/file\/9ef2ed4b7fd2c810847ffa5fa85bce38-Paper.pdf"},{"key":"10233_CR27","doi-asserted-by":"publisher","DOI":"10.1017\/9781108571401","volume-title":"Bandit Algorithms","author":"T Lattimore","year":"2020","unstructured":"Lattimore, T., Szepesv\u00e1ri, C.: Bandit Algorithms. Cambridge University Press, Cambridge (2020)"},{"key":"10233_CR28","unstructured":"LeCun, Y., Cortes, C.: MNIST handwritten digit database (2010). http:\/\/yann.lecun.com\/exdb\/mnist\/"},{"issue":"2","key":"10233_CR29","doi-asserted-by":"publisher","first-page":"A712","DOI":"10.1137\/15M102318X","volume":"38","author":"B Leimkuhkler","year":"2016","unstructured":"Leimkuhkler, B., Xiaocheng, S.: Adaptive thermostats for noisy gradient systems. SIAM J. Sci. Comput. 38(2), A712\u2013A736 (2016)","journal-title":"SIAM J. Sci. Comput."},{"key":"10233_CR30","doi-asserted-by":"crossref","unstructured":"Li, C., Chen, C., Carlson, D., et\u00a0al. Preconditioned stochastic gradient Langevin dynamics for deep neural networks. In: Proceedings of the AAAI Conference on Artificial Intelligence (2016)","DOI":"10.1609\/aaai.v30i1.10200"},{"key":"10233_CR31","unstructured":"Liu, Q., Lee, J., Jordan, M.: A kernelized stein discrepancy for goodness-of-fit tests. In: International Conference on Machine Learning, pp. 276\u2013284 (2016)"},{"key":"10233_CR32","unstructured":"Ma, Y.A., Chen, T., Fox, E.: A complete recipe for stochastic gradient MCMC. In: Advances in Neural Information Processing Systems, pp. 2917\u20132925 (2015)"},{"key":"10233_CR33","unstructured":"Nemeth, C., Fearnhead, P.: Stochastic gradient Markov chain Monte Carlo. J. Am. Stat. Assoc., pp. 1\u201318 (2020)"},{"key":"10233_CR34","unstructured":"Phan, D., Pradhan, N., Jankowiak, M.: Composable effects for flexible and accelerated probabilistic programming in numpyro. (2019) arXiv preprint arXiv:1912.11554"},{"issue":"6","key":"10233_CR35","doi-asserted-by":"publisher","first-page":"2320","DOI":"10.1214\/11-AAP828","volume":"22","author":"NS Pillai","year":"2012","unstructured":"Pillai, N.S., Stuart, A.M., Thi\u00e9ry, A.H., et al.: Optimal scaling and diffusion limits for the Langevin algorithm in high dimensions. Ann. Appl. Probab. 22(6), 2320\u20132356 (2012)","journal-title":"Ann. Appl. Probab."},{"issue":"1","key":"10233_CR36","doi-asserted-by":"publisher","first-page":"255","DOI":"10.1111\/1467-9868.00123","volume":"60","author":"GO Roberts","year":"1998","unstructured":"Roberts, G.O., Rosenthal, J.S.: Optimal scaling of discrete approximations to Langevin diffusions. J. R. Stat. Soc.: Ser. B (Stat. Methodol.) 60(1), 255\u2013268 (1998)","journal-title":"J. R. Stat. Soc.: Ser. B (Stat. Methodol.)"},{"issue":"4","key":"10233_CR37","doi-asserted-by":"publisher","first-page":"341","DOI":"10.2307\/3318418","volume":"2","author":"GO Roberts","year":"1996","unstructured":"Roberts, G.O., Tweedie, R.L.: Exponential convergence of Langevin distributions and their discrete approximations. Bernoulli 2(4), 341\u2013363 (1996)","journal-title":"Bernoulli"},{"key":"10233_CR38","doi-asserted-by":"crossref","unstructured":"Salakhutdinov, R., Mnih, A.: Bayesian probabilistic matrix factorization using Markov chain Monte Carlo. In: Proceedings of the 25th International Conference on Machine Learning, ACM, pp. 880\u2013887 (2008)","DOI":"10.1145\/1390156.1390267"},{"key":"10233_CR39","volume-title":"Approximation Theorems of Mathematical Statistics","author":"RJ Serfling","year":"2009","unstructured":"Serfling, R.J.: Approximation Theorems of Mathematical Statistics, vol. 162. Wiley, London (2009)"},{"key":"10233_CR40","doi-asserted-by":"crossref","unstructured":"Slivkins, A.: Introduction to multi-armed bandits (2019). arXiv preprint arXiv:1904.07272","DOI":"10.1561\/9781680836219"},{"issue":"5","key":"10233_CR41","doi-asserted-by":"publisher","first-page":"997","DOI":"10.1007\/s11222-011-9269-5","volume":"22","author":"M Vihola","year":"2012","unstructured":"Vihola, M.: Robust adaptive Metropolis algorithm with coerced acceptance rate. Stat. Comput. 22(5), 997\u20131008 (2012)","journal-title":"Stat. Comput."},{"key":"10233_CR42","unstructured":"Welling, M., Teh, Y.W.: Bayesian learning via stochastic gradient Langevin dynamics. In: Proceedings of the 28th International Conference on Machine Learning (ICML), pp. 681\u2013688 (2011)"}],"container-title":["Statistics and Computing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s11222-023-10233-3.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s11222-023-10233-3\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s11222-023-10233-3.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,5,7]],"date-time":"2023-05-07T00:05:10Z","timestamp":1683417910000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s11222-023-10233-3"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,4,4]]},"references-count":42,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2023,6]]}},"alternative-id":["10233"],"URL":"https:\/\/doi.org\/10.1007\/s11222-023-10233-3","relation":{"has-preprint":[{"id-type":"doi","id":"10.21203\/rs.3.rs-1914735\/v1","asserted-by":"object"}]},"ISSN":["0960-3174","1573-1375"],"issn-type":[{"value":"0960-3174","type":"print"},{"value":"1573-1375","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,4,4]]},"assertion":[{"value":"31 July 2022","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"5 March 2023","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"4 April 2023","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors have no conflicts of interest to disclose. LFS acknowledges financial support from the Queensland University of Technology Centre for Data Science. JC was supported by EPSRC grant EP\/S00159X\/1 and CN acknowledges the support of EPSRC grants EP\/R01860X\/1, EP\/S00159X\/1 and EP\/V022636\/1.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflicts of interest"}}],"article-number":"66"}}