Synonyms
Definition
A Markov Chain Monte Carlo (MCMC) algorithm is a method for sequential sampling in which each new sample is drawn from the neighborhood of its predecessor. This sequence forms a Markov chain, since the transition probabilities between sample values are only dependent on the last sample value. MCMC algorithms are well suited to sampling in high-dimensional spaces.
Motivation
Sampling from a probability density function is necessary in many kinds of approximation, including Bayesian inference and other applications in Machine Learning. However, sampling is not always easy, especially in high-dimensional spaces. Mackay (2003) gives a simple example to illustrate the problem. Suppose we want to find the average concentration of plankton in a lake, whose profile looks like this:
If we do not know the depth profile of the lake, how would we know where to sample from? If we take a boat out, would we have to sample almost exhaustively by fixing a grid on the surface of...
Recommended Reading
MCMC is well covered in several text books. Mackay (2003) gives a thorough and readable introduction to MCMC and Gibbs Sampling. Russell (2009) explain MCMC in the context of approximate inference for Bayesian networks. Hastie et al. (2009) also give a more technical account of sampling from the posterior. Andrieu et al. (2003) Machine Learning paper gives a thorough introduction to MCMC for Machine Learning. There are also some excellent tutorials on the web including Walsh (2004) and Iain Murray’s video tutorial (Murray, 2009) for machine learning summer school.
Andrieu, C., DeFreitas, N., Doucet, A., & Jordan, M. I. (2003). An introduction to MCMC for machine learning. Machine Learning, 50(1), 5–43.
Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: data mining, inference and perception (2nd ed.). New York: Springer.
Hastings, W. K. (1970). Monte Carlo sampling methods using Markov chains and their applications. Biometrika, 57, 97–109.
Mackay, D. J. C. (2003). Information theory, inference and learning algorithms. Cambridge: Cambridge University Press.
Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, A., & Teller, H. (1953). Equations of state calculations by fast computing machines. Journal of Chemical Physics, 21, 1087–1091.
Metropolis, N., & Ulam, S. (1949). The Monte Carlo method. Journal of the American Statistical Association, 44(247), 335–341.
Murray, I. (2009). Markov chain Monte Carlo. http://videolectures.net/mlss09uk_murray_mcmc/. Retrieved 25 July 2010.
Russell, S., & Norvig, P. (2009). Artificial intelligence: a modern approach (3rd ed.). NJ: Prentice Hall.
Walsh, B. (2004). Markov chain Monte Carlo and Gibbs sampling. http://nitro.biosci.arizona.edu/courses/EEB581-2004/handouts/Gibbs. Retrieved 25 July 2010.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2011 Springer Science+Business Media, LLC
About this entry
Cite this entry
Sammut, C. (2011). Markov Chain Monte Carlo. In: Sammut, C., Webb, G.I. (eds) Encyclopedia of Machine Learning. Springer, Boston, MA. https://doi.org/10.1007/978-0-387-30164-8_511
Download citation
DOI: https://doi.org/10.1007/978-0-387-30164-8_511
Publisher Name: Springer, Boston, MA
Print ISBN: 978-0-387-30768-8
Online ISBN: 978-0-387-30164-8
eBook Packages: Computer ScienceReference Module Computer Science and Engineering