Advanced Computational Methods for Bayesian Signal Processing

Страна: N/A

Город: N/A

Тезисы до: 30.09.2016

Даты: 30.09.16 — 30.09.16

Область наук: Технические;

Е-мейл Оргкомитета:

Организаторы: EURASIP Journal on Advances in Signal Processing w


The problem of estimating some variables of interest from noisy observations is ubiquitous in different fields, such as signal processing, finance, oceanography, video tracking and so on. Computational methods are often required in Bayesian inference and nonlinear signal processing to deal with intractable posterior densities. For instance, Sequential Importance Sampling (a.k.a. particle filters) and Markov Chain Monte Carlo (MCMC) methods, which have been popular approaches within the statistical community for a long time, have been widely used in signal processing and communications applications. Over the last years, several extensions and variants of these two families of methods have been proposed in order to improve their performance (e.g., for the estimation of fixed parameters or dealing with multi-modal target densities): population Monte Carlo (PMC) schemes, particle MCMC (PMCMC), adaptive Monte Carlo approaches (i.e., MCMC with adaptive proposal functions), multiple try Metropolis (MTM) strategies, parallel Monte Carlo chains, etc. Some of these methods have found their way into the signal processing literature, but there are still many recent advanced Monte Carlo methods, developed within the statistical community, that are not so widely known by signal processing practitioners and which may be very useful for signal processing applications. This special issue intends to bridge the gap between both communities by presenting a collection of papers that describe recent advances in Monte Carlo methods with signal processing applications in mind.

Throughout the years, the problems addressed in the field of statistical signal processing have become increasingly challenging. On the one hand, the probabilistic models have become more complex in an attempt to better capture the specific characteristics of the different applications. On the other hand, despite the availability of parallelized computing architectures and of always cheaper data storage, increasing flows of high dimensional data are well known to remain a major obstacle which impacts the applicability of available solutions. Altogether, these two facts imply that analytical solutions are impossible to compute in most practical applications, and thus efficient computational methods are called for. Unfortunately, many advanced computational methods, which have been recently developed within the field of Bayesian inference, are still not widely applied by signal processing practitioners. The main goal of this special issue is to introduce novel approaches from the statistical community to a wider signal processing audience. The focus will be both on the theoretical/methodological aspects (introducing recently developed algorithms) and their applications, especially within the field of Bayesian signal processing.
Potential topics include, but are not limited to:

    Sequential importance sampling (a.k.a. particle filters)
    SMC theory (convergence property, numerical stability)
    Sequential (or not) MC within SMC
    SMC for multi-target and/or multi-sensor filtering
    SMC based on evolutionary strategies
    Smoothing problem, path space SMC
    Quasi Monte Carlo (QMC) and Sequential Quasi Monte Carlo (SQMC) methods
    Particle MCMC (PMCMC) and Marginal MCMC
    Adaptive Monte Carlo approaches (MCMC with adaptive proposal functions, SMC samplers, Adaptive importance sampling (AIS), population Monte Carlo (PMC) schemes...)
    Data assimilation algorithms, Mixed SMC/Ensemble Kalman filter techniques
    Multiple try Metropolis (MTM) strategies
    Interacting Parallel Monte Carlo chains and island particle filters
    Computational methods for models with intractable likelihood
    Approximate Bayesian computation (ABC)
    Bayesian signal processing applications: localization and tracking, SLAM algorithms, model selection, parameter estimation in state space models, big data, etc.

Веб-сайт конференции:

Конференции по теме - с близкими дедлайнами: