{"type":"ScholarlyArticle","title":"An Efficient Minibatch Acceptance Test for Metropolis-Hastings","authors":[{"name":"Daniel Seita","slug":"daniel-seita","given":"Daniel","family":"Seita","affiliation":"University of California , Berkeley , CA"},{"name":"Xinlei Pan","slug":"xinlei-pan","given":"Xinlei","family":"Pan","affiliation":"University of California , Berkeley , CA"},{"name":"Haoyu Chen","slug":"haoyu-chen","given":"Haoyu","family":"Chen","affiliation":"University of California , Berkeley , CA"},{"name":"John Canny","slug":"john-canny","given":"John","family":"Canny","affiliation":"University of California , Berkeley , CA; Google Research , Mountain View , CA"}],"abstract":"We present a novel Metropolis-Hastings method for large datasets that uses small expected-size mini-batches of data. Previous work on reducing the cost of Metropolis-Hastings tests yields only constant factor reductions versus using the full dataset for each sample. Here we present a method that can be tuned to provide arbitrarily small batch sizes, by adjusting either proposal step size or temperature. Our test uses the noise-tolerant Barker acceptance test with a novel additive correction variable. The resulting test has similar cost to a normal SGD update. Our experiments demonstrate several order-of-magnitude speedups over previous work.","doi":"10.24963/ijcai.2018/753","url":"https://ijcai.org/proceedings/2018/753","pdf":"https://ijcai.org/proceedings/2018/0753.pdf","markdown":"https://ijcai.org/proceedings/2018/753.md","bibtex":"https://ijcai.org/proceedings/2018/bibtex/753","volume":{"slug":"2018","title":"Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2018","url":"https://ijcai.org/proceedings/2018/","year":2018,"location":"Stockholm","dates":"13-19 July 2018","editors":"Jérôme Lang","isbn":"978-0-9992411-2-7","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2018"},"section":"Sister Conferences Best Papers","track":"Best Sister Conferences","number":753,"pages":{"first":5359,"last":5363},"keywords":["Uncertainty in AI: Approximate Probabilistic Inference","Machine Learning: Machine Learning"],"videos":[],"references":[{"position":1,"raw":"[Ahn et al., 2012] Sungjin Ahn, Anoop Korattikara Balan, and Max Welling. Bayesian Posterior Sampling via Stochastic Gradient Fisher Scoring. In International Con- ference on Machine Learning (ICML), 2012.","title":"Bayesian Posterior Sampling via Stochastic Gradient Fisher Scoring","authors":"Anoop Sungjin Ahn, Max Korattikara Balan, Welling","year":"2012","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[Bardenet et al., 2014] Rémi Bardenet, Arnaud Doucet, and Chris Holmes. Towards Scaling up Markov chain Monte Carlo: An Adaptive Subsampling Approach. In Interna- tional Conference on Machine Learning (ICML), 2014.","title":"Towards Scaling up Markov chain Monte Carlo: An Adaptive Subsampling Approach","authors":"Rémi Bardenet, Arnaud Doucet, Chris Holmes","year":"2014","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Bardenet et al., 2017] Rémi Bardenet, Arnaud Doucet, and Chris Holmes. On Markov Chain Monte Carlo Methods for Tall Data. Journal of Machine Learning Research (JMLR), 2017.","title":"On Markov Chain Monte Carlo Methods for Tall Data","authors":"Rémi Bardenet, Arnaud Doucet, Chris Holmes","year":"2017","venue":"Journal of Machine Learning Research","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Barker, 1965] A. A. Barker. Monte-Carlo Calculations of the Radial Distribution Functions for a Proton-Electron Plasma. Australian Journal of Physics, 18:119-133, 1965.","title":"Monte-Carlo Calculations of the Radial Distribution Functions for a Proton-Electron Plasma","authors":"A A Barker","year":"1965","venue":"Australian Journal of Physics","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Betancourt, 2017] Michael Betancourt. A Conceptual In- troduction to Hamiltonian Monte Carlo. arXiv preprint arXiv:1701.02434, 2017.","title":"A Conceptual Introduction to Hamiltonian Monte Carlo","authors":"Michael Betancourt","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Brooks et al., 2011] Steve Brooks, Andrew Gelman, Galin Jones, and Xiao-Li Meng. Handbook of Markov Chain Monte Carlo. CRC press, 2011.","title":"Handbook of Markov Chain Monte Carlo","authors":"Steve Brooks, Andrew Gelman, Galin Jones, Xiao-Li Meng","year":"2011","venue":null,"doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Chen and Ghahramani, 2016] Yutian Chen and Zoubin Ghahramani. Scalable Discrete Sampling as a Multi- Armed Bandit Problem. In International Conference on Machine Learning (ICML), 2016.","title":"Scalable Discrete Sampling as a Multi-Armed Bandit Problem","authors":"Chen, Ghahramani ; ] Yutian Chen, Zoubin Ghahramani","year":"2016","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Chen et al., 2014] T. Chen, E.B. Fox, and C. Guestrin. Stochastic Gradient Hamiltonian Monte Carlo. In Inter- national Conference on Machine Learning (ICML), 2014.","title":"Stochastic Gradient Hamiltonian Monte Carlo","authors":"T Chen, E B Fox, C Guestrin","year":"2014","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Dupuy and Bach, 2016] Christophe Dupuy and Francis Bach. Online but Accurate Inference for Latent Vari- able Models with Local Gibbs Sampling. arXiv preprint arXiv:1603.02644, 2016.","title":"Online but Accurate Inference for Latent Variable Models with Local Gibbs Sampling","authors":"Christophe Dupuy, Francis Bach","year":"2016","venue":null,"doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Gilks and Spiegelhalter, 1996] W.R. Gilks and DJ Spiegel- halter. Markov chain Monte Carlo in practice. Chapman & Hall/CRC, 1996.","title":"Markov chain Monte Carlo in practice","authors":"W R Gilks, Spiegelhalter","year":"1996","venue":null,"doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Hastings, 1970] W. K. Hastings. Monte Carlo Sampling Methods Using Markov Chains and Their Applications. Biometrika, 57:97-109, 1970.","title":"Monte Carlo Sampling Methods Using Markov Chains and Their Applications","authors":"W K Hastings","year":"1970","venue":"Biometrika","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Homan and Gelman, 2014] Matthew D. Homan and An- drew Gelman. The No-U-turn Sampler: Adaptively Set- ting Path Lengths in Hamiltonian Monte Carlo. Journal of Machine Learning Research (JMLR), 15(1):1593-1623, January 2014.","title":"The No-U-turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo","authors":"Matthew D Homan, Andrew Gelman","year":"2014","venue":"Journal of Machine Learning Research (JMLR)","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Korattikara et al., 2014] Anoop Korattikara, Yutian Chen, and Max Welling. Austerity in MCMC Land: Cutting the Metropolis-Hastings Budget. In International Conference on Machine Learning (ICML), 2014.","title":"Austerity in MCMC Land: Cutting the Metropolis-Hastings Budget","authors":"Anoop Korattikara, Yutian Chen, Max Welling","year":"2014","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[LeCun and Cortes, 1998] Yann LeCun and Corinna Cortes. MNIST Handwritten Digit Database. 1998.","title":"MNIST Handwritten Digit Database","authors":"Lecun, Cortes, Yann Lecun, Corinna Cortes","year":"1998","venue":null,"doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Loosli et al., 2007] Gaëlle Loosli, Stéphane Canu, and Léon Bottou. Training Invariant Support Vector Machines us- ing Selective Sampling. In Léon Bottou, Olivier Chapelle, Dennis DeCoste, and Jason Weston, editors, Large Scale Kernel Machines, pages 301-320. MIT Press, Cambridge, MA., 2007.","title":"Training Invariant Support Vector Machines using Selective Sampling","authors":"Gaëlle Loosli, Stéphane Canu, Léon Bottou","year":"2007","venue":"Large Scale Kernel Machines","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Maclaurin and Adams, 2014] Dougal Maclaurin and Ryan P. Adams. Firefly Monte Carlo: Exact MCMC with Subsets of Data. In Conference on Uncertainty in Artificial Intelligence (UAI), 2014.","title":"Firefly Monte Carlo: Exact MCMC with Subsets of Data","authors":"Dougal Maclaurin, Ryan P Adams","year":"2014","venue":"Conference on Uncertainty in Artificial Intelligence (UAI)","doi":null,"url":null,"internal_url":"/proceedings/2015/610"},{"position":17,"raw":"[Metropolis et al., 1953] Nicholas Metropolis, Arianna W. Rosenbluth, Marshall N. Rosenbluth, Augusta H. Teller, and Edward Teller. Equation of State Calculations by Fast Computing Machines. The Journal of Chemical Physics, 21, 1953.","title":"Equation of State Calculations by Fast Computing Machines","authors":"Nicholas Metropolis, Arianna W Rosenbluth, Marshall N Rosenbluth, Augusta H Teller, Edward Teller","year":"1953","venue":"The Journal of Chemical Physics","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Neal, 2010] Radford M. Neal. MCMC Using Hamiltonian Dynamics. Handbook of Markov Chain Monte Carlo, 54:113-162, 2010.","title":"MCMC Using Hamiltonian Dynamics","authors":"M Radford, Neal","year":"2010","venue":"Handbook of Markov Chain Monte Carlo","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Novak, 2005] Y. Novak. On Self-Normalized Sums and Student's Statistic. Theory of Probability and its Appli- cations, 49(2):336-344, 2005.","title":"On Self-Normalized Sums and Student's Statistic","authors":"Y Novak","year":"2005","venue":"Theory of Probability and its Applications","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Roberts and Rosenthal, 2001] Gareth O. Roberts and Jef- frey S. Rosenthal. Optimal Scaling for Various Metropolis-Hastings Algorithms. Statistical Science, 16(4):351-367, 2001.","title":"Optimal Scaling for Various Metropolis-Hastings Algorithms","authors":"O Gareth, Jeffrey S Roberts, Rosenthal","year":"2001","venue":"Statistical Science","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Seita et al., 2017] Daniel Seita, Xinlei Pan, Haoyu Chen, and John Canny. An Efficient Minibatch Acceptance Test for Metropolis Hastings. In Conference on Uncertainty in Artificial Intelligence (UAI), 2017.","title":"An Efficient Minibatch Acceptance Test for Metropolis Hastings","authors":"Daniel Seita, Xinlei Pan, Haoyu Chen, John Canny","year":"2017","venue":"Conference on Uncertainty in Artificial Intelligence (UAI)","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Tripuraneni et al., 2017] Nilesh Tripuraneni, Mark Row- land, Zoubin Ghahramani, and Richard Turner. Magnetic Hamiltonian Monte Carlo. In International Conference on Machine Learning (ICML), 2017.","title":"Magnetic Hamiltonian Monte Carlo","authors":"Nilesh Tripuraneni, Mark Rowland, Zoubin Ghahramani, Richard Turner","year":"2017","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Wang and Uhlenbeck, 1945] M.C. Wang and G.E. Uhlen- beck. On the Theory of the Brownian Motion II. Reviews of Modern Physics, 1945.","title":"On the Theory of the Brownian Motion II","authors":"M C Wang, G E Uhlenbeck","year":"1945","venue":"Reviews of Modern Physics","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Welling and Teh, 2011] Max Welling and Yee Whye Teh. Bayesian Learning via Stochastic Gradient Langevin Dy- namics. In International Conference on Machine Learning (ICML), 2011.","title":"Bayesian Learning via Stochastic Gradient Langevin Dynamics","authors":"Teh Welling, Max Welling, Yee Whye Teh","year":"2011","venue":"International Conference on Machine Learning (ICML)","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}