Pseudo-Marginal Inference for CTMCs on Infinite Spaces via Monotonic Likelihood Approximations
From MaRDI portal
Publication:6094080
Abstract: Bayesian inference for Continuous-Time Markov Chains (CTMCs) on countably infinite spaces is notoriously difficult because evaluating the likelihood exactly is intractable. One way to address this challenge is to first build a non-negative and unbiased estimate of the likelihood -- involving the matrix exponential of finite truncations of the true rate matrix -- and then to use the estimates in a pseudo-marginal inference method. In this work, we show that we can dramatically increase the efficiency of this approach by avoiding the computation of exact matrix exponentials. In particular, we develop a general methodology for constructing an unbiased, non-negative estimate of the likelihood using doubly-monotone matrix exponential approximations. We further develop a novel approximation in this family -- the skeletoid -- as well as theory regarding its approximation error and how that relates to the variance of the estimates used in pseudo-marginal inference. Experimental results show that our approach yields more efficient posterior inference for a wide variety of CTMCs.
Cites work
- scientific article; zbMATH DE number 6378055 (Why is no real title available?)
- scientific article; zbMATH DE number 3082073 (Why is no real title available?)
- A general method for debiasing a Monte Carlo estimator
- A robust discrete state approximation to the optimal nonlinear filter for a diffusiont
- Algorithm 955: Approximation of the inverse Poisson cumulative distribution function
- An Exact Gibbs Sampler for the Markov-Modulated Poisson Process
- Auxiliary Variable Methods for Markov Chain Monte Carlo with Applications
- Bayesian analysis of elapsed times in continuous‐time Markov chains
- Closed-form likelihood expansions for multivariate diffusions
- Computing the action of the matrix exponential, with an application to exponential integrators
- Concentration inequalities. A nonasymptotic theory of independence
- Continuous-time Markov chains. An applications-oriented approach
- Continuous-time approximations for the nonlinear filtering problem
- Direct statistical inference for finite Markov jump processes via the matrix exponential
- Efficient Parameter Sampling for Markov Jump Processes
- Efficient parameter estimation for DNA kinetics modeled as continuous-time Markov chains
- Exact Inference for Continuous Time Markov Chain Models
- Exact and Computationally Efficient Likelihood-Based Estimation for Discretely Observed Diffusion Processes (with Discussion)
- Large-sample asymptotics of the pseudo-marginal method
- Markov chain Monte Carlo for exact inference for diffusions
- Markov models for accumulating mutations
- Maximum-likelihood estimation for diffusion processes via closed-form density expansions
- Nineteen Dubious Ways to Compute the Exponential of a Matrix, Twenty-Five Years Later
- On Russian roulette estimates for Bayesian inference with doubly-intractable likelihoods
- On nonnegative unbiased estimators
- Stochastic simulation of the kinetics of multiple interacting nucleic acid strands
- The linear birth-death process: an inferential retrospective
- The pseudo-marginal approach for efficient Monte Carlo computations
- Unbiased Bayesian inference for population Markov jump processes via random truncations
- Unbiased estimation with square root convergence for SDE models
This page was built for publication: Pseudo-Marginal Inference for CTMCs on Infinite Spaces via Monotonic Likelihood Approximations
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6094080)
