Simon J. Godsill

On sequential Monte Carlo sampling methods for Bayesian filtering

In this article, we present an overview of methods for sequential simulation from posterior distributions. These methods are of particular interest in Bayesian filtering for discrete time dynamic models that are typically nonlinear and non-Gaussian. A general importance sampling framework is developed that unifies many of the methods which have been proposed over the last few decades in several different scientific disciplines. Novel extensions to the existing methods are also proposed. We show in particular how to incorporate local linearisation methods similar to those which have previously been employed in the deterministic filtering literature; these lead to very effective importance distributions. Furthermore we describe a method which uses Rao-Blackwellisation in order to take advantage of the analytic structure present in some important classes of state-space models. In a final section we develop algorithms for prediction, smoothing and evaluation of the likelihood in dynamic models.

An Overview of Existing Methods and Recent Advances in Sequential Monte Carlo

It is now over a decade since the pioneering contribution of Gordon (1993), which is commonly regarded as the first instance of modern sequential Monte Carlo (SMC) approaches. Initially focussed on applications to tracking and vision, these techniques are now very widespread and have had a significant impact in virtually all areas of signal and image processing concerned with Bayesian dynamical models. This paper is intended to serve both as an introduction to SMC algorithms for nonspecialists and as a reference to recent contributions in domains where the techniques are still under significant development, including smoothing, estimation of fixed parameters and use of SMC methods beyond the standard filtering contexts.

Monte Carlo Smoothing for Nonlinear Time Series

We develop methods for performing smoothing computations in general state-space models. The methods rely on a particle representation of the filtering distributions, and their evolution through time using sequential importance sampling and resampling ideas. In particular, novel techniques are presented for generation of sample realizations of historical state sequences. This is carried out in a forward-filtering backward-smoothing procedure that can be viewed as the nonlinear, non-Gaussian counterpart of standard Kalman filter-based simulation smoothers in the linear Gaussian case. Convergence in the mean squared error sense of the smoothed trajectories is proved, showing the validity of our proposed method. The methods are tested in a substantial application for the processing of speech signals represented by a time-varying autoregression and parameterized in terms of time-varying partial correlation coefficients, comparing the results of our algorithm with those from a simple smoother based on the filtered trajectories.

Monte Carlo filtering for multi target tracking and data association

We present Monte Carlo methods for multi-target tracking and data association. The methods are applicable to general nonlinear and non-Gaussian models for the target dynamics and measurement likelihood. We provide efficient solutions to two very pertinent problems: the data association problem that arises due to unlabelled measurements in the presence of clutter, and the curse of dimensionality that arises due to the increased size of the state-space associated with multiple targets. We develop a number of algorithms to achieve this. The first, which we refer to as the Monte Carlo joint probabilistic data association filter (MC-JPDAF), is a generalisation of the strategy proposed by Schulz et al. (2001) and Schulz et al. (2003). As is the case for the JPDAF, the distributions of interest are the marginal filtering distributions for each of the targets, but these are approximated with particles rather than Gaussians. We also develop two extensions to the standard particle filtering methodology for tracking multiple targets. The first, which we refer to as the sequential sampling particle filter (SSPF), samples the individual targets sequentially by utilising a factorisation of the importance weights. The second, which we refer to as the independent partition particle filter (IPPF), assumes the associations to be independent over the individual targets, leading to an efficient component-wise sampling strategy to construct new particles. We evaluate and compare the proposed methods on a challenging synthetic tracking problem.

On the Relationship Between Markov chain Monte Carlo Methods for Model Uncertainty

This article considers Markov chain computational methods for incorporating uncertainty about the dimension of a parameter when performing inference within a Bayesian setting. A general class of methods is proposed for performing such computations, based upon a product space representation of the problem which is similar to that of Carlin and Chib. It is shown that all of the existing algorithms for incorporation of model uncertainty into Markov chain Monte Carlo (MCMC) can be derived as special cases of this general class of methods. In particular, we show that the popular reversible jump method is obtained when a special form of Metropolis–Hastings (M–H) algorithm is applied to the product space. Furthermore, the Gibbs sampling method and the variable selection method are shown to derive straightforwardly from the general framework. We believe that these new relationships between methods, which were until now seen as diverse procedures, are an important aid to the understanding of MCMC model selection procedures and may assist in the future development of improved procedures. Our discussion also sheds some light upon the important issues of “pseudo-prior” selection in the case of the Carlin and Chib sampler and choice of proposal distribution in the case of reversible jump. Finally, we propose efficient reversible jump proposal schemes that take advantage of any analytic structure that may be present in the model. These proposal schemes are compared with a standard reversible jump scheme for the problem of model order uncertainty in autoregressive time series, demonstrating the improvements which can be achieved through careful choice of proposals.

A Bayesian Approach for Blind Separation of Sparse Sources

We present a Bayesian approach for blind separation of linear instantaneous mixtures of sources having a sparse representation in a given basis. The distributions of the coefficients of the sources in the basis are modeled by a Student t distribution, which can be expressed as a scale mixture of Gaussians, and a Gibbs sampler is derived to estimate the sources, the mixing matrix, the input noise variance and also the hyperparameters of the Student t distributions. The method allows for separation of underdetermined (more sources than sensors) noisy mixtures. Results are presented with audio signals using a modified discrete cosine transform basis and compared with a finite mixture of Gaussians prior approach. These results show the improved sound quality obtained with the Student t prior and the better robustness to mixing matrices close to singularity of the Markov chain Monte Carlo approach

Particle methods for Bayesian modeling and enhancement of speech signals

This paper applies time-varying autoregressive (TVAR) models with stochastically evolving parameters to the problem of speech modeling and enhancement. The stochastic evolution models for the TVAR parameters are Markovian diffusion processes. The main aim of the paper is to perform on-line estimation of the clean speech and model parameters and to determine the adequacy of the chosen statistical models. Efficient particle methods are developed to solve the optimal filtering and fixed-lag smoothing problems. The algorithms combine sequential importance sampling (SIS), a selection step and Markov chain Monte Carlo (MCMC) methods. They employ several variance reduction strategies to make the best use of the statistical structure of the model. It is also shown how model adequacy may be determined by combining the particle filter with frequentist methods. The modeling and enhancement performance of the models and estimation algorithms are evaluated in simulation studies on both synthetic and real speech data sets.

Efficient alternatives to the Ephraim and Malah suppression rule for audio signal enhancement

Audio signal enhancement often involves the application of a time-varying filter, or suppression rule, to the frequency-domain transform of a corrupted signal. Here we address suppression rules derived under a Gaussian model and interpret them as spectral estimators in a Bayesian statistical framework. With regard to the optimal spectral amplitude estimator of Ephraim and Malah, we show that under the same modelling assumptions, alternative methods of Bayesian estimation lead to much simpler suppression rules exhibiting similarly effective behaviour. We derive three of such rules and demonstrate that, in addition to permitting a more straightforward implementation, they yield a more intuitive interpretation of the Ephraim and Malah solution.

Maximum a Posteriori Sequence Estimation Using Monte Carlo Particle Filters

We develop methods for performing maximum a posteriori (MAP) sequence estimation in non-linear non-Gaussian dynamic models. The methods rely on a particle cloud representation of the filtering distribution which evolves through time using importance sampling and resampling ideas. MAP sequence estimation is then performed using a classical dynamic programming technique applied to the discretised version of the state space. In contrast with standard approaches to the problem which essentially compare only the trajectories generated directly during the filtering stage, our method efficiently computes the optimal trajectory over all combinations of the filtered states. A particular strength of the method is that MAP sequence estimation is performed sequentially in one single forwards pass through the data without the requirement of an additional backward sweep. An application to estimation of a non-linear time series model and to spectral estimation for time-varying autoregressions is described.

Bayesian harmonic models for musical pitch estimation and analysis

Estimating the pitch of musical signals is complicated by the presence of partials in addition to the fundamental frequency. In this paper, we propose developments to an earlier Bayesian model which describes each component signal in terms of fundamental frequency, partials (‘harmonics’), and amplitude. This basic model is modified for greater realism to include non-white residual spectrum, time-varying amplitudes and partials ‘detuned’ from the natural linear relationship. The unknown parameters of the new model are simulated using a reversible jump MCMC algorithm, leading to a highly accurate pitch estimator. The models and algorithms can be applied for feature extraction, polyphonic music transcription, source separation and restoration of musical sources.