Abstract
We introduce an adaptive output-sensitive Metropolis-Hastings algorithm for probabilistic models expressed as programs, Adaptive Lightweight Metropolis-Hastings (AdLMH). This algorithm extends Lightweight Metropolis-Hastings (LMH) by adjusting the probabilities of proposing random variables for modification to improve convergence of the program output. We show that AdLMH converges to the correct equilibrium distribution and compare convergence of AdLMH to that of LMH on several test problems to highlight different aspects of the adaptation scheme. We observe consistent improvement in convergence on the test problems.
Chapter PDF
Similar content being viewed by others
References
Andrieu, C., Thoms, J.: A tutorial on adaptive MCMC. Statistics and Computing 18(4), 343–373 (2008)
Atchadé, Y., Fort, G., Moulines, E., Priouret, P.: Adaptive markov chain monte carlo: theory and methods. In: Barber, D., Cemgil, A.T., Chiappa, S. (eds.) Bayesian Time Series Models, pp. 32–51. Cambridge University Press (2011)
Auer, P., Cesa-Bianchi, N., Fischer, P.: Finite-time analysis of the Multiarmed Bandit problem. Machine Learning 47(2–3), 235–256 (2002)
Bai, Y., Roberts, G.O., Rosenthal, J.S.: On the containment condition for adaptive Markov chain Monte Carlo algorithms. Advances and Applications in Statistics 21(1), 1–54 (2011)
Gamerman, D., Lopes, H.F.: Markov Chain Monte Carlo: Stochastic Simulation for Bayesian Inference. Chapman and Hall/CRC (2006)
Goodman, N.D., Mansinghka, V.K., Roy, D.M., Bonawitz, K., Tenenbaum, J.B.: Church: a language for generative models. In: UAI, pp. 220–229 (2008)
Gordon, A.D., Henzinger, T.A., Nori, A.V., Rajamani, S.K.: Probabilistic programming. In: ICSE (FOSE track) (2014)
Łatuszyński, K., Roberts, G.O., Rosenthal, J.S.: Adaptive Gibbs samplers and related MCMC methods. Annals of Applied Probability 23(1), 66–98 (2013)
Lauritzen, S.: Graphical Models. Clarendon Press (1996)
Levine, R.A., Yu, Z., Hanley, W.G., Nitao, J.J.: Implementing componentwise hastings algorithms. Computational Stastistics & Data Analysis 48(2), 363–389 (2005)
Mansinghka, V.K., Selsam, D., Perov, Y.N.: Venture: a higher-order probabilistic programming platform with programmable inference. CoRR abs/1404.0099 (2014)
van de Meent, J.W., Yang, H., Mansinghka, V., Wood, F.: Particle Gibbs with Ancestor Sampling for Probabilistic Programs. In: AISTATS, pp. 986–994 (2015)
Nori, A.V., Hur, C.K., Rajamani, S.K., Samuel, S.: R2: An efficient mcmc sampler for probabilistic programs. In: AAAI, pp. 2476–2482 (2014)
Roberts, G.O., Rosenthal, J.S.: Coupling and ergodicity of adaptive MCMC. Journal of Applied Probability 44, 458–475 (2007)
Roberts, G.O., Rosenthal, J.S.: Examples of adaptive MCMC. Journal of Computational and Graphical Statistics 18(2), 349–367 (2009)
Wingate, D., Stuhlmüller, A., Goodman, N.D.: Lightweight implementations of probabilistic programming languages via transformational compilation. In: AISTATS, pp. 770–778 (2011)
Wood, F., van de Meent, J.W., Mansinghka, V.: A new approach to probabilistic programming inference. In: AISTATS, pp. 1024–1032 (2014)
Yang, L., Hanrahan, P., Goodman, N.D.: Generating efficient MCMC kernels from probabilistic programs. In: AISTATS, pp. 1068–1076 (2014)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2015 Springer International Publishing Switzerland
About this paper
Cite this paper
Tolpin, D., van de Meent, JW., Paige, B., Wood, F. (2015). Output-Sensitive Adaptive Metropolis-Hastings for Probabilistic Programs. In: Appice, A., Rodrigues, P., Santos Costa, V., Gama, J., Jorge, A., Soares, C. (eds) Machine Learning and Knowledge Discovery in Databases. ECML PKDD 2015. Lecture Notes in Computer Science(), vol 9285. Springer, Cham. https://doi.org/10.1007/978-3-319-23525-7_19
Download citation
DOI: https://doi.org/10.1007/978-3-319-23525-7_19
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-23524-0
Online ISBN: 978-3-319-23525-7
eBook Packages: Computer ScienceComputer Science (R0)