Searcharxiv⌕ Search

arXiv subjects

Arnaud Doucet

Publications and source records attributed to Arnaud Doucet.

At least 163 records · Page 9Linked to original sources
arXiv2015

On Particle Methods for Parameter Estimation in State-Space Models

Nonlinear non-Gaussian state-space models are ubiquitous in statistics, econometrics, information engineering and signal processing. Particle methods, also known as Sequential Monte Carlo (SMC) methods, provide reliable numerical approximations to the associated state inference problems. However, in most applications, the state-space model of interest also depends on unknown static parameters that need to be estimated from the data. In this context, standard particle methods fail and it is necessary to rely on more sophisticated algorithms. The aim of this paper is to present a comprehensive review of particle methods that have been proposed to perform static parameter estimation in state-space models. We discuss the advantages and limitations of these methods and illustrate their performance on simple models.

stat.CO↗
arXiv2015

Derivative-Free Estimation of the Score Vector and Observed Information Matrix with Application to State-Space Models

Ionides, King et al. (see e.g. Inference for nonlinear dynamical systems, PNAS 103) have recently introduced an original approach to perform maximum likelihood parameter estimation in state-space models which only requires being able to simulate the latent Markov model according to its prior distribution. Their methodology relies on an approximation of the score vector for general statistical models based upon an artificial posterior distribution and bypasses the calculation of any derivative. We show here that this score estimator can be derived from a simple application of Stein's lemma and how an additional application of this lemma provides an original derivative-free estimator of the observed information matrix. We establish that these estimators exhibit robustness properties compared to finite difference estimators while their bias and variance scale as well as finite difference type estimators, including simultaneous perturbations (see e.g. Spall, IEEE Trans. on Automatic Control 37), with respect to the dimension of the parameter. For state-space models where sequential Monte Carlo computation is required, these estimators can be further improved. In this specific context, we derive original derivative-free estimators of the score vector and observed information matrix which are computed using sequential Monte Carlo approximations of smoothed additive functionals associated with a modified version of the original state-space model.

stat.ME↗
arXiv2015

Expectation Particle Belief Propagation

We propose an original particle-based implementation of the Loopy Belief Propagation (LPB) algorithm for pairwise Markov Random Fields (MRF) on a continuous state space. The algorithm constructs adaptively efficient proposal distributions approximating the local beliefs at each note of the MRF. This is achieved by considering proposal distributions in the exponential family whose parameters are updated iterately in an Expectation Propagation (EP) framework. The proposed particle scheme provides consistent estimation of the LBP marginals as the number of particles increases. We demonstrate that it provides more accurate results than the Particle Belief Propagation (PBP) algorithm of Ihler and McAllester (2009) at a fraction of the computational cost and is additionally more robust empirically. The computational complexity of our algorithm at each iteration is quadratic in the number of particles. We also propose an accelerated implementation with sub-quadratic computational complexity which still provides consistent estimates of the loopy BP marginal distributions and performs almost as well as the original procedure.

stat.CO↗
arXiv2015

On Markov chain Monte Carlo methods for tall data

Markov chain Monte Carlo methods are often deemed too computationally intensive to be of any practical use for big data applications, and in particular for inference on datasets containing a large number $n$ of individual data points, also known as tall datasets. In scenarios where data are assumed independent, various approaches to scale up the Metropolis-Hastings algorithm in a Bayesian inference context have been recently proposed in machine learning and computational statistics. These approaches can be grouped into two categories: divide-and-conquer approaches and, subsampling-based algorithms. The aims of this article are as follows. First, we present a comprehensive review of the existing literature, commenting on the underlying assumptions and theoretical guarantees of each method. Second, by leveraging our understanding of these limitations, we propose an original subsampling-based approach which samples from a distribution provably close to the posterior distribution of interest, yet can require less than $O(n)$ data point likelihood evaluations at each iteration for certain statistical models in favourable scenarios. Finally, we have only been able so far to propose subsampling-based methods which display good performance in scenarios where the Bernstein-von Mises approximation of the target posterior distribution is excellent. It remains an open challenge to develop such methods in scenarios where the Bernstein-von Mises approximation is poor.

stat.ME↗
arXiv2014

Efficient implementation of Markov chain Monte Carlo when using an unbiased likelihood estimator

When an unbiased estimator of the likelihood is used within a Metropolis--Hastings chain, it is necessary to trade off the number of Monte Carlo samples used to construct this estimator against the asymptotic variances of averages computed under this chain. Many Monte Carlo samples will typically result in Metropolis--Hastings averages with lower asymptotic variances than the corresponding Metropolis--Hastings averages using fewer samples. However, the computing time required to construct the likelihood estimator increases with the number of Monte Carlo samples. Under the assumption that the distribution of the additive noise introduced by the log-likelihood estimator is Gaussian with variance inversely proportional to the number of Monte Carlo samples and independent of the parameter value at which it is evaluated, we provide guidelines on the number of samples to select. We demonstrate our results by considering a stochastic volatility model applied to stock index returns.

stat.ME↗
arXiv2014

Perfect simulation using atomic regeneration with application to Sequential Monte Carlo

Consider an irreducible, Harris recurrent Markov chain of transition kernel Π and invariant probability measure π. If Π satisfies a minorization condition, then the split chain allows the identification of regeneration times which may be exploited to obtain perfect samples from π. Unfortunately, many transition kernels associated with complex Markov chain Monte Carlo algorithms are analytically intractable, so establishing a minorization condition and simulating the split chain is challenging, if not impossible. For uniformly ergodic Markov chains with intractable transition kernels, we propose two efficient perfect simulation procedures of similar expected running time which are instances of the multigamma coupler and an imputation scheme. These algorithms overcome the intractability of the kernel by introducing an artificial atom and using a Bernoulli factory. We detail an application of these procedures when Π is the recently introduced iterated conditional Sequential Monte Carlo kernel. We additionally provide results on the general applicability of the methodology, and how Sequential Monte Carlo methods may be used to facilitate perfect simulation and/or unbiased estimation of expectations with respect to the stationary distribution of a non-uniformly ergodic Markov chain.

stat.CO↗
arXiv2014

Asynchronous Anytime Sequential Monte Carlo

We introduce a new sequential Monte Carlo algorithm we call the particle cascade. The particle cascade is an asynchronous, anytime alternative to traditional particle filtering algorithms. It uses no barrier synchronizations which leads to improved particle throughput and memory efficiency. It is an anytime algorithm in the sense that it can be run forever to emit an unbounded number of particles while keeping within a fixed memory budget. We prove that the particle cascade is an unbiased marginal likelihood estimator which means that it can be straightforwardly plugged into existing pseudomarginal methods.

stat.CO↗
arXiv2014

Fast Computation of Wasserstein Barycenters

We present new algorithms to compute the mean of a set of empirical probability measures under the optimal transport metric. This mean, known as the Wasserstein barycenter, is the measure that minimizes the sum of its Wasserstein distances to each element in that set. We propose two original algorithms to compute Wasserstein barycenters that build upon the subgradient method. A direct implementation of these algorithms is, however, too costly because it would require the repeated resolution of large primal and dual optimal transport problems to compute subgradients. Extending the work of Cuturi (2013), we propose to smooth the Wasserstein distance used in the definition of Wasserstein barycenters with an entropic regularizer and recover in doing so a strictly convex objective whose gradients can be computed for a considerably cheaper computational cost using matrix scaling algorithms. We use these algorithms to visualize a large family of images and to solve a constrained clustering problem.

stat.ML↗
arXiv2013

A Lognormal Central Limit Theorem for Particle Approximations of Normalizing Constants

This paper deals with the numerical approximation of normalizing constants produced by particle methods, in the general framework of Feynman-Kac sequences of measures. It is well-known that the corresponding estimates satisfy a central limit theorem for a fixed time horizon $n$ as the number of particles $N$ goes to infinity. Here, we study the situation where both $n$ and $N$ go to infinity in such a way that $\lim_{n\rightarrow\infty}% n/N=α>0$. In this context, Pitt et al. \cite{pitt2012} recently conjectured that a lognormal central limit theorem should hold. We formally establish this result here, under general regularity assumptions on the model. We also discuss special classes of models (time-homogeneous environment and ergodic random environment) for which more explicit descriptions of the limiting bias and variance can be obtained.

math.PR↗
arXiv2013

Reversible Jump MCMC Simulated Annealing for Neural Networks

We propose a novel reversible jump Markov chain Monte Carlo (MCMC) simulated annealing algorithm to optimize radial basis function (RBF) networks. This algorithm enables us to maximize the joint posterior distribution of the network parameters and the number of basis functions. It performs a global search in the joint space of the parameters and number of parameters, thereby surmounting the problem of local minima. We also show that by calibrating a Bayesian model, we can obtain the classical AIC, BIC and MDL model selection criteria within a penalized likelihood framework. Finally, we show theoretically and empirically that the algorithm converges to the modes of the full posterior distribution in an efficient way.

cs.LG↗
arXiv2013

Rao-Blackwellised Particle Filtering for Dynamic Bayesian Networks

Particle filters (PFs) are powerful sampling-based inference/learning algorithms for dynamic Bayesian networks (DBNs). They allow us to treat, in a principled way, any type of probability distribution, nonlinearity and non-stationarity. They have appeared in several fields under such names as "condensation", "sequential Monte Carlo" and "survival of the fittest". In this paper, we show how we can exploit the structure of the DBN to increase the efficiency of particle filtering, using a technique known as Rao-Blackwellisation. Essentially, this samples some of the variables, and marginalizes out the rest exactly, using the Kalman filter, HMM filter, junction tree algorithm, or any other finite dimensional optimal filter. We show that Rao-Blackwellised particle filters (RBPFs) lead to more accurate estimates than standard PFs. We demonstrate RBPFs on two problems, namely non-stationary online regression with radial basis function networks and robot localization and map building. We also discuss other potential application areas and provide references to some finite dimensional optimal filters.

cs.LG↗
arXiv2012

Sequentially interacting Markov chain Monte Carlo methods

Sequential Monte Carlo (SMC) is a methodology for sampling approximately from a sequence of probability distributions of increasing dimension and estimating their normalizing constants. We propose here an alternative methodology named Sequentially Interacting Markov Chain Monte Carlo (SIMCMC). SIMCMC methods work by generating interacting non-Markovian sequences which behave asymptotically like independent Metropolis-Hastings (MH) Markov chains with the desired limiting distributions. Contrary to SMC, SIMCMC allows us to iteratively improve our estimates in an MCMC-like fashion. We establish convergence results under realistic verifiable assumptions and demonstrate its performance on several examples arising in Bayesian time series analysis.

math.ST↗
arXiv2012

Toward Practical N2 Monte Carlo: the Marginal Particle Filter

Sequential Monte Carlo techniques are useful for state estimation in non-linear, non-Gaussian dynamic models. These methods allow us to approximate the joint posterior distribution using sequential importance sampling. In this framework, the dimension of the target distribution grows with each time step, thus it is necessary to introduce some resampling steps to ensure that the estimates provided by the algorithm have a reasonable variance. In many applications, we are only interested in the marginal filtering distribution which is defined on a space of fixed dimension. We present a Sequential Monte Carlo algorithm called the Marginal Particle Filter which operates directly on the marginal distribution, hence avoiding having to perform importance sampling on a space of growing dimension. Using this idea, we also derive an improved version of the auxiliary particle filter. We show theoretic and empirical results which demonstrate a reduction in variance over conventional particle filtering, and present techniques for reducing the cost of the marginal particle filter with N particles from O(N2) to O(N logN).

stat.CO↗
arXiv2012

Generalized Polya Urn for Time-varying Dirichlet Process Mixtures

Dirichlet Process Mixtures (DPMs) are a popular class of statistical models to perform density estimation and clustering. However, when the data available have a distribution evolving over time, such models are inadequate. We introduce here a class of time-varying DPMs which ensures that at each time step the random distribution follows a DPM model. Our model relies on an intuitive and simple generalized Polya urn scheme. Inference is performed using Markov chain Monte Carlo and Sequential Monte Carlo. We demonstrate our model on various applications.

stat.ME↗
arXiv2012

Distributed Maximum Likelihood for Simultaneous Self-localization and Tracking in Sensor Networks

We show that the sensor self-localization problem can be cast as a static parameter estimation problem for Hidden Markov Models and we implement fully decentralized versions of the Recursive Maximum Likelihood and on-line Expectation-Maximization algorithms to localize the sensor network simultaneously with target tracking. For linear Gaussian models, our algorithms can be implemented exactly using a distributed version of the Kalman filter and a novel message passing algorithm. The latter allows each node to compute the local derivatives of the likelihood or the sufficient statistics needed for Expectation-Maximization. In the non-linear case, a solution based on local linearization in the spirit of the Extended Kalman Filter is proposed. In numerical examples we demonstrate that the developed algorithms are able to learn the localization parameters.

math.OC↗
arXiv2012

An Adaptive Interacting Wang-Landau Algorithm for Automatic Density Exploration

While statisticians are well-accustomed to performing exploratory analysis in the modeling stage of an analysis, the notion of conducting preliminary general-purpose exploratory analysis in the Monte Carlo stage (or more generally, the model-fitting stage) of an analysis is an area which we feel deserves much further attention. Towards this aim, this paper proposes a general-purpose algorithm for automatic density exploration. The proposed exploration algorithm combines and expands upon components from various adaptive Markov chain Monte Carlo methods, with the Wang-Landau algorithm at its heart. Additionally, the algorithm is run on interacting parallel chains -- a feature which both decreases computational cost as well as stabilizes the algorithm, improving its ability to explore the density. Performance is studied in several applications. Through a Bayesian variable selection example, the authors demonstrate the convergence gains obtained with interacting chains. The ability of the algorithm's adaptive proposal to induce mode-jumping is illustrated through a trimodal density and a Bayesian mixture modeling application. Lastly, through a 2D Ising model, the authors demonstrate the ability of the algorithm to overcome the high correlations encountered in spatial models.

stat.CO↗
arXiv2012

New inference strategies for solving Markov Decision Processes using reversible jump MCMC

In this paper we build on previous work which uses inferences techniques, in particular Markov Chain Monte Carlo (MCMC) methods, to solve parameterized control problems. We propose a number of modifications in order to make this approach more practical in general, higher-dimensional spaces. We first introduce a new target distribution which is able to incorporate more reward information from sampled trajectories. We also show how to break strong correlations between the policy parameters and sampled trajectories in order to sample more freely. Finally, we show how to incorporate these techniques in a principled manner to obtain estimates of the optimal policy.

cs.LG↗