Adaptive sequential Monte Carlo by means of mixture of experts

DOI10.1007/S11222-012-9372-2zbMATH Open1325.62151arXiv1108.2836OpenAlexW1978093868MaRDI QIDQ892475FDOQ892475

Authors: Julien Cornebise, Eric Moulines, Jimmy Olsson 

Publication date: 19 November 2015

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: Appropriately designing the proposal kernel of particle filters is an issue of significant importance, since a bad choice may lead to deterioration of the particle sample and, consequently, waste of computational power. In this paper we introduce a novel algorithm adaptively approximating the so-called optimal proposal kernel by a mixture of integrated curved exponential distributions with logistic weights. This family of distributions, referred to as mixtures of experts, is broad enough to be used in the presence of multi-modality or strongly skewed distributions. The mixtures are fitted, via online-EM methods, to the optimal kernel through minimisation of the Kullback-Leibler divergence between the auxiliary target and instrumental distributions of the particle filter. At each iteration of the particle filter, the algorithm is required to solve only a single optimisation problem for the whole particle sample, yielding an algorithm with only linear complexity. In addition, we illustrate in a simulation study how the method can be successfully applied to optimal filtering in nonlinear state-space models.

Full work available at URL: https://arxiv.org/abs/1108.2836

Recommendations

• An adaptive sequential Monte Carlo method for approximate Bayesian computation
• An adaptive sequential Monte Carlo sampler
• Sequential Monte Carlo with adaptive weights for approximate Bayesian computation
• Adaptive Incremental Mixture Markov Chain Monte Carlo
• On the convergence of adaptive sequential Monte Carlo methods
• Adaptive multilevel Monte Carlo for probabilities
• On adaptive Markov chain Monte Carlo algorithms
• Sequential Monte Carlo methods for mixtures with normalized random measures with independent increments priors
• On adaptive resampling strategies for sequential Monte Carlo methods
• An adaptive approach to Langevin MCMC

zbMATH Keywords

sequential Monte Carloparticle filterKullback-Leibler divergenceShannon entropyadaptive algorithmscoefficient of variationexpectation-maximisationoptimal proposal kernel

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Statistical aspects of information-theoretic topics (62B10) Markov processes: estimation; hidden Markov models (62M05) Monte Carlo methods (65C05) Sequential statistical analysis (62L10)

Cites Work

• Sequential Monte Carlo Methods in Practice
• Sequential Monte Carlo Samplers
• A New Approach to Estimating Switching Regressions
• An adaptive Metropolis algorithm
• Inference in hidden Markov models.
• The EM Algorithm and Extensions, 2E
• Title not available (Why is that?)
• Filtering via Simulation: Auxiliary Particle Filters
• Monte Carlo strategies in scientific computing
• Title not available (Why is that?)
• On the ergodicity properties of some adaptive MCMC algorithms
• On-line expectation-maximization algorithm for latent data models
• Particle Markov chain Monte Carlo for efficient numerical simulation
• Title not available (Why is that?)
• On adaptive resampling strategies for sequential Monte Carlo methods
• On the auxiliary particle filter
• The transform likelihood ratio method for rare event simulation with heavy tails
• Integration of Multimodal Functions by Monte Carlo Importance Sampling
• Covariance pooling and stabilization for classification

Cited In (4)

• Importance sampling from posterior distributions using copula-like approximations
• Antithetic sampling for sequential Monte Carlo methods with application to state-space models
• Sequential Monte Carlo with kernel embedded mappings: the mapping particle filter
• Title not available (Why is that?)

This page was built for publication: Adaptive sequential Monte Carlo by means of mixture of experts

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q892475)