{"type":"ScholarlyArticle","title":"Variance Reduction in Black-box Variational Inference by Adaptive Importance Sampling","authors":[{"name":"Ximing Li","slug":"ximing-li","given":"Ximing","family":"Li","affiliation":"College of Computer Science and Technology , Jilin University , China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education , China"},{"name":"Changchun Li","slug":"changchun-li","given":"Changchun","family":"Li","affiliation":"College of Computer Science and Technology , Jilin University , China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education , China"},{"name":"Jinjin Chi","slug":"jinjin-chi","given":"Jinjin","family":"Chi","affiliation":"College of Computer Science and Technology , Jilin University , China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education , China"},{"name":"Jihong Ouyang","slug":"jihong-ouyang","given":"Jihong","family":"Ouyang","affiliation":"College of Computer Science and Technology , Jilin University , China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education , China"}],"abstract":"Overdispersed black-box variational inference employs importance sampling to reduce the variance of the Monte Carlo gradient in black-box variational inference. A simple overdispersed proposal distribution is used. This paper aims to investigate how to adaptively obtain better proposal distribution for lower variance. To this end, we directly approximate the optimal proposal in theory using a Monte Carlo moment matching step at each variational iteration. We call this adaptive proposal moment matching proposal (MMP). Experimental results on two Bayesian models show that the MMP can effectively reduce variance in black-box learning, and perform better than baseline inference algorithms.","doi":"10.24963/ijcai.2018/333","url":"https://ijcai.org/proceedings/2018/333","pdf":"https://ijcai.org/proceedings/2018/0333.pdf","markdown":"https://ijcai.org/proceedings/2018/333.md","bibtex":"https://ijcai.org/proceedings/2018/bibtex/333","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":"Machine Learning","track":"Main track","number":333,"pages":{"first":2404,"last":2410},"keywords":["Machine Learning: Learning Generative Models","Machine Learning: Probabilistic Machine Learning"],"videos":[],"references":[{"position":1,"raw":"[Bishop, 2006] Christopher M. Bishop. Pattern Recognition and Machine Learning. Springer, New York, 2006.","title":"Pattern Recognition and Machine Learning","authors":"Christopher M Bishop","year":"2006","venue":null,"doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[Blei and Lafferty, 2007] David M. Blei and John D. Laffer- ty. A correlated topic model of science. Annals of Applied Statistics, 1(1):17-35, 2007.","title":"A correlated topic model of science","authors":"David M Blei, John D Lafferty","year":"2007","venue":"Annals of Applied Statistics","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Blei et al., 2003] David M. Blei, Andrew Y. Ng, and Michael I. Jordan. Latent Dirichlet allocation. Journal of Machine Learning Research, 3:993-1022, 2003.","title":"Latent Dirichlet allocation","authors":"David M Blei, Andrew Y Ng, Michael I Jordan","year":"2003","venue":"Journal of Machine Learning Research","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Burda et al., 2016] Yuri Burda, Roger Grosse, and Ruslan Salakhutdinov. Importance weighted autoencoders. Inter- national Conference on Learning Representations, 2016.","title":"Importance weighted autoencoders","authors":"Yuri Burda, Roger Grosse, Ruslan Salakhutdinov","year":"2016","venue":"International Conference on Learning Representations","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Ghahramani and Beal, 2001] Zoubin Ghahramani and Matthew J. Beal. Propagation algorithms for variational Bayesian learning. Neural Information Processing Systems, 2001.","title":"Propagation algorithms for variational Bayesian learning","authors":"Beal Ghahramani, Zoubin Ghahramani, Matthew J Beal","year":"2001","venue":"Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Ghahramani and Jordan, 1997] Zoubin Ghahramani and Michael I. Jordan. Factorial hidden markov models. Machine Learning, 29(2-3):245-273, 1997.","title":"Factorial hidden markov models","authors":"Zoubin Ghahramani, Michael I Jordan","year":"1997","venue":"Machine Learning","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Honkela and Valpola, 2004] Antti Honkela and Harri Valpola. Unsupervised variational bayesian learning of nonlinear models. Neural Information Processing Systems, 2004.","title":"Unsupervised variational bayesian learning of nonlinear models","authors":"Valpola Honkela, Antti Honkela, Harri Valpola","year":"2004","venue":"Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Jaakkola and Jordan, 1997] Tommi S. Jaakkola and Michael I. Jordan. A variational approach to bayesian logistic regression models and their extensions. Artificial Intelligence and Statistics, 1997.","title":"A variational approach to bayesian logistic regression models and their extensions","authors":"Tommi S Jaakkola, Michael I Jordan","year":"1997","venue":"Artificial Intelligence and Statistics","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Jang et al., 2017] Eric Jang, Shixiang Gu, and Ben Poole. Categorical reparameterization using gumbel-softmax. In- ternational Conference on Learning Representations, 2017.","title":"Categorical reparameterization using gumbel-softmax","authors":"Eric Jang, Shixiang Gu, Ben Poole","year":"2017","venue":"ternational Conference on Learning Representations","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Jordan, 1999] Michael I. Jordan. Learning in Graphical Models. MIT Press, Cambridge, MA, 1999.","title":"Learning in Graphical Models","authors":"I Michael, Jordan","year":"1999","venue":null,"doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Jorgensen, 1987] Bent Jorgensen. Exponential dispersion models. Journal of the Royal Statistical Society, Series B (Methodological), 49(2):127-162, 1987.","title":"Bent Jorgensen. Exponential dispersion models","authors":"Jorgensen","year":"1987","venue":"Journal of the Royal Statistical Society, Series B (Methodological)","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Kingma and Welling, 2014] Durk Kingma and Max Welling. Auto-encoding variational Bayes. International Conference on Learning Representations, 2014.","title":"Auto-encoding variational Bayes","authors":"Welling Kingma, Durk Kingma, Max Welling","year":"2014","venue":"International Conference on Learning Representations","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Knowles, 2015] David A. Knowles. Stochastic gradien- t variational bayes for Gamma approximating distribution- s. arXiv:1509.01631, 2015.","title":"Stochastic gradient variational bayes for Gamma approximating distributions","authors":"David A Knowles","year":"2015","venue":null,"doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Maddison et al., 2017] Chris J. Maddison, Andriy Mnih, and Yee Whye Teh. The concrete distribution: A continu- ous relaxation of discrete random variables. International Conference on Learning Representations, 2017.","title":"The concrete distribution: A continuous relaxation of discrete random variables","authors":"Chris J Maddison, Andriy Mnih, Yee Whye Teh","year":"2017","venue":"International Conference on Learning Representations","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Minka and Lafferty, 2002] Thomas P Minka and John Laf- ferty. Expectation-propagation for the generative aspect model. Conference on Uncertainty in Artificial Intelligenc, pages 352-359, 2002.","title":"Expectation-propagation for the generative aspect model","authors":"P Thomas, John Minka, Lafferty","year":"2002","venue":"Conference on Uncertainty in Artificial Intelligenc","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Minka, 2001] Thomas P Minka. Expectation propagation for approximate Bayesian inference. Conference on Un- certainty in Artificial Intelligenc, pages 362-369, 2001.","title":"Expectation propagation for approximate Bayesian inference","authors":"Thomas P Minka","year":"2001","venue":"Conference on Uncertainty in Artificial Intelligenc","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Minka, 2005] Thomas P Minka. Divergence measures and message passing. (MSR-TR-2005-173), 2005.","title":"Divergence measures and message passing","authors":"Thomas P Minka","year":"2005","venue":null,"doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Mnih and Rezende, 2016] Andriy Mnih and Danilo J. Rezende. Variational inference for monte carlo objective. International Conference on Machine Learning, 2016.","title":"Variational inference for monte carlo objective","authors":"Andriy Mnih, Danilo J Rezende","year":"2016","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Naesseth et al., 2017] Christian A. Naesseth, Francisco J. R. Ruiz, Scott W. Lindermanz, and David M. Blei. Reparameterization gradients through acceptance- rejection sampling algorithms. International Conference on Artificial Intelligence and Statistics, 2017.","title":"Reparameterization gradients through acceptancerejection sampling algorithms","authors":"Christian A Naesseth, J R Francisco, Scott W Ruiz, David M Lindermanz, Blei","year":"2017","venue":"International Conference on Artificial Intelligence and Statistics","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Owen, 2013] Art B. Owen. Monte Carlo theory, method- s and examples. http://statweb.stanford.edu/∼owen/mc/, 2013.","title":"Monte Carlo theory, methods and examples","authors":"Art B Owen, Owen","year":"2013","venue":null,"doi":null,"url":"http://statweb.stanford.edu/∼owen/mc/,2013","internal_url":null},{"position":21,"raw":"[Paisley et al., 2012] John Paisley, David M. Blei, and Michael I. Jordan. Variational Bayesian inference with stochastic search. International Conference on Machine Learning, 2012.","title":"Variational Bayesian inference with stochastic search","authors":"John Paisley, David M Blei, Michael I Jordan","year":"2012","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Papadimitriou and Steiglitz, 1998] Christos H Papadimitri- ou and Kenneth Steiglitz. Combinatorial optimization: al- gorithms and complexity. Courier Corporation, 1998.","title":"Combinatorial optimization: algorithms and complexity","authors":"Christos H Papadimitriou, Kenneth Steiglitz","year":"1998","venue":null,"doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Ranganath et al., 2014] Rajesh Ranganath, Sean Gerrish, and David M. Blei. Black-box variational inference. Arti- ficial Intelligence and Statistics, 2014.","title":"Artificial Intelligence and Statistics","authors":"Rajesh Ranganath, Sean Gerrish, David M Blei","year":"2014","venue":null,"doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Robbins and Monro, 1951] Herbert Robbins and Sutton Monro. A stochastic approximation method. The Annals of Mathematical Statistics, 22(3):400-407, 1951.","title":"Herbert Robbins and Sutton Monro. A stochastic approximation method","authors":"Monro Robbins","year":"1951","venue":"The Annals of Mathematical Statistics","doi":null,"url":null,"internal_url":null},{"position":25,"raw":"[Rubinstein and Kroese, 2016] Reuven Y. Rubinstein and Dirk P. Kroese. Simulation and the Monte Carlo Method, 3rd Edition. Wiley Series in Probability and Statistics, 2016.","title":"Simulation and the Monte Carlo Method","authors":"Reuven Y Rubinstein, Dirk P Kroese","year":"2016","venue":"Wiley Series in Probability and Statistics","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"[Ruiz et al., 2016a] Francisco J. R. Ruiz, Michalis K. Titsias, and David M. Blei. The generalized reparameterization gradient. Neural Information Processing Systems, 2016.","title":"The generalized reparameterization gradient","authors":"Ruiz","year":"2016","venue":"Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":27,"raw":"[Ruiz et al., 2016b] Francisco J. R. Ruiz, Michalis K. Tit- sias, and David M. Blei. Overdispersed black-box vari- ational inference. Conference on Uncertainty in Artificial Intelligence, pages 647-656, 2016.","title":"Overdispersed black-box variational inference","authors":"Ruiz","year":"2016","venue":"Conference on Uncertainty in Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":28,"raw":"[Titsias and Lazaro-Gredilla, 2015] Michalis K. Titsias and Miguel Lazaro-Gredilla. local expectation gradients for black box variational inference. Neural Information Pro- cessing Systems, pages 2620-2628, 2015.","title":"local expectation gradients for black box variational inference","authors":"K Michalis, Miguel Titsias, Lazaro-Gredilla","year":"2015","venue":"Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":29,"raw":"[Wainwright and Jordan, 2008] Martin J. Wainwright and Michael I. Jordan. Graphical models, exponential fami- lies, and variational inference. Foundations and Trends in Machine Learning, 1(1-2):1-305, 2008.","title":"Graphical models, exponential families, and variational inference","authors":"Martin J Wainwright, Michael I Jordan","year":"2008","venue":"Foundations and Trends in Machine Learning","doi":null,"url":null,"internal_url":null},{"position":30,"raw":"[Wang and Blei, 2013] Chong Wang and David M. Blei. Variational inference in nonconjugate models. Journal of Machine Learning Research, 14:1005-1031, 2013.","title":"Variational inference in nonconjugate models","authors":"Chong Wang, David M Blei","year":"2013","venue":"Journal of Machine Learning Research","doi":null,"url":null,"internal_url":null},{"position":31,"raw":"[Williams, 1992] Ronald J. Williams. Simple statistical gra- dient following algorithms for connectionist reinforce- ment learning. Machine Learning, 8(3-4):229-256, 1992.","title":"Simple statistical gradient following algorithms for connectionist reinforcement learning","authors":"Ronald J Williams","year":"1992","venue":"Machine Learning","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}