Monte Carlo inference via greedy importance sampling
نویسندگان
چکیده
We present a new method for conducting Monte Carlo inference in graphical models which combines explicit search with generalized importance sampling. The idea is to reduce the variance of importance sampling by searching for significant points in the target distribution. We prove that it is possible to introduce search and still maintain unbiasedness. We then demonstrate our procedure on a few simple inference tasks and show that it can improve the inference quality of standard MCMC methods, including Gibbs sampling, Metropolis sampling, and Hybrid Monte Carlo. This paper extends previous work which showed how greedy importance sampling could be correctly realized in the one-dimensional case.
منابع مشابه
Greedy importance sampling: A new Monte Carlo inference method
This paper presents a new Monte Carlo inference method that is a simple variation of importance sampling. It is well known that importance sampling fails when the proposal distribution concentrates samples in low density regions of the target distribution, which leads to high variance in the estimates. The method we propose attempts to mitigate this difficulty by explicitly searching for high d...
متن کاملMonte Carlo Techniques for Bayesian Statistical Inference – A comparative review
In this article, we summariseMonte Carlo simulationmethods commonly used in Bayesian statistical computing. We give descriptions for each algorithm and provide R codes for their implementation via a simple 2-dimensional example. We compare the relative merits of these methods qualitatively by considering their general user-friendliness, and numerically in terms of mean squared error and computa...
متن کاملRegularized Greedy Importance Sampling
Greedy importance sampling is an unbiased estimation technique that reduces the variance of standard importance sampling by explicitly searching for modes in the estimation objective. Previous work has demonstrated the feasibility of implementing this method and proved that the technique is unbiased in both discrete and continuous domains. In this paper we present a reformulation of greedy impo...
متن کاملGreedy Importance Sampling
I present a simple variation of importance sampling that explicitly searches for important regions in the target distribution. I prove that the technique yields unbiased estimates, and show empirically it can reduce the variance of standard Monte Carlo estimators. This is achieved by concentrating samples in more significant regions of the sample space.
متن کاملSupplemental Information Likelihood-based inference in isolation-by-distance models using the spatial distribution of low-frequency alleles
To integrate over Tl we use straightforward Monte Carlo sampling, i.e. for the ith Monte Carlo replicate we sample the vector ti from the distribution P (Tl). To integrate over Gl we use importance sampling, so that for the ith Monte Carlo replicate we sample gi from an importance sampling distribution P (Gl) and weight its contribution by w(gi). We describe both of these steps in more detail b...
متن کامل