Efficient implementation of Markov chain Monte Carlo when using an unbiased likelihood estimator

Duration: 34 mins 45 secs

Share this media item:

Embed this media item:

<iframe width="_width_" height="_height_" src="https://sms.cam.ac.uk/media/1704624/embed" frameborder="0" scrolling="no" allowfullscreen></iframe>

Choose size:

About this item

Available Formats

About this item

Description:	Pitt, M (University of Warwick) Tuesday 22 April 2014, 13:45-14:20


Created:	2014-04-25 13:40
Collection:	Advanced Monte Carlo Methods for Complex Inference Problems
Publisher:	Isaac Newton Institute
Copyright:	Pitt, M
Language:	eng (English)


Abstract:	When an unbiased estimator of the likelihood is used within an Metropolis-Hastings scheme, it is necessary to tradeoff the number of samples used to evaluate the likelihood against the computing time. Many samples will result in a scheme which has similar properties to the case where the likelihood is exactly known but will be expensive. Few samples will result in faster estimation but at the expense of slower mixing of the Markov chain. We explore the relationship between the number of samples and the efficiency of the resulting Metropolis-Hastings estimates. Under the assumption that the distribution of the additive noise introduced by the log-likelihood estimator is independent of the point at which this log-likelihood is evaluated and other relatively mild assumptions, we provide guidelines on the number of samples to select for a general Metropolis-Hastings proposal. We illustrate on a complex stochastic volatility model that these assumptions are approximately satisfied experimentally and that the theoretical insights with regards to inefficiency and computational time hold true. Keywords: Bayesian inference; Estimated likelihood; Metropolis-Hastings; Particle filtering.

Abstract:

When an unbiased estimator of the likelihood is used within an Metropolis-Hastings scheme, it is necessary to tradeoff the number of samples used to evaluate the likelihood against the computing time. Many samples will result in a scheme which has similar properties to the case where the likelihood is exactly known but will be expensive. Few samples will result in faster estimation but at the expense of slower mixing of the Markov chain. We explore the relationship between the number of samples and the efficiency of the resulting Metropolis-Hastings estimates. Under the assumption that the distribution of the additive noise introduced by the log-likelihood estimator is independent of the point at which this log-likelihood is evaluated and other relatively mild assumptions, we provide guidelines on the number of samples to select for a general Metropolis-Hastings proposal. We illustrate on a complex stochastic volatility model that these assumptions are approximately satisfied experimentally and that the theoretical insights with regards to inefficiency and computational time hold true.
Keywords: Bayesian inference; Estimated likelihood; Metropolis-Hastings; Particle filtering.

Available Formats

Format	Quality	Bitrate	Size
MPEG-4 Video	640x360	1.94 Mbits/sec	505.94 MB	View	Download
WebM	640x360	480.16 kbits/sec	122.21 MB	View	Download
iPod Video	480x270	521.94 kbits/sec	132.84 MB	View	Download
MP3	44100 Hz	249.74 kbits/sec	63.63 MB	Listen	Download
Auto *	(Allows browser to choose a format it supports)