Involutive MCMC: a Unifying Framework

Kirill Neklyudov, Max Welling, Evgenii Egorov, Dmitry Vetrov
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:7273-7282, 2020.

Abstract

Markov Chain Monte Carlo (MCMC) is a computational approach to fundamental problems such as inference, integration, optimization, and simulation. The field has developed a broad spectrum of algorithms, varying in the way they are motivated, the way they are applied and how efficiently they sample. Despite all the differences, many of them share the same core principle, which we unify as the Involutive MCMC (iMCMC) framework. Building upon this, we describe a wide range of MCMC algorithms in terms of iMCMC, and formulate a number of “tricks” which one can use as design principles for developing new MCMC algorithms. Thus, iMCMC provides a unified view of many known MCMC algorithms, which facilitates the derivation of powerful extensions. We demonstrate the latter with two examples where we transform known reversible MCMC algorithms into more efficient irreversible ones.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-neklyudov20a, title = {Involutive {MCMC}: a Unifying Framework}, author = {Neklyudov, Kirill and Welling, Max and Egorov, Evgenii and Vetrov, Dmitry}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {7273--7282}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/neklyudov20a/neklyudov20a.pdf}, url = {https://proceedings.mlr.press/v119/neklyudov20a.html}, abstract = {Markov Chain Monte Carlo (MCMC) is a computational approach to fundamental problems such as inference, integration, optimization, and simulation. The field has developed a broad spectrum of algorithms, varying in the way they are motivated, the way they are applied and how efficiently they sample. Despite all the differences, many of them share the same core principle, which we unify as the Involutive MCMC (iMCMC) framework. Building upon this, we describe a wide range of MCMC algorithms in terms of iMCMC, and formulate a number of “tricks” which one can use as design principles for developing new MCMC algorithms. Thus, iMCMC provides a unified view of many known MCMC algorithms, which facilitates the derivation of powerful extensions. We demonstrate the latter with two examples where we transform known reversible MCMC algorithms into more efficient irreversible ones.} }
Endnote
%0 Conference Paper %T Involutive MCMC: a Unifying Framework %A Kirill Neklyudov %A Max Welling %A Evgenii Egorov %A Dmitry Vetrov %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-neklyudov20a %I PMLR %P 7273--7282 %U https://proceedings.mlr.press/v119/neklyudov20a.html %V 119 %X Markov Chain Monte Carlo (MCMC) is a computational approach to fundamental problems such as inference, integration, optimization, and simulation. The field has developed a broad spectrum of algorithms, varying in the way they are motivated, the way they are applied and how efficiently they sample. Despite all the differences, many of them share the same core principle, which we unify as the Involutive MCMC (iMCMC) framework. Building upon this, we describe a wide range of MCMC algorithms in terms of iMCMC, and formulate a number of “tricks” which one can use as design principles for developing new MCMC algorithms. Thus, iMCMC provides a unified view of many known MCMC algorithms, which facilitates the derivation of powerful extensions. We demonstrate the latter with two examples where we transform known reversible MCMC algorithms into more efficient irreversible ones.
APA
Neklyudov, K., Welling, M., Egorov, E. & Vetrov, D.. (2020). Involutive MCMC: a Unifying Framework. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:7273-7282 Available from https://proceedings.mlr.press/v119/neklyudov20a.html.

Related Material