Increasingly, for many application areas, it is becoming important to include elements of nonlinearity and non-Gaussianity in order to model accurately the underlying dynamics of a physical system. Moreover, it is typically crucial to process data on-line as it arrives, both from the point of view of storage costs as well as for rapid adaptation to changing signal characteristics. In this paper, we review both optimal and suboptimal Bayesian algorithms for nonlinear/non-Gaussian tracking problems, with a focus on particle filters. Particle filters are sequential Monte Carlo methods based on point mass (or "particle") representations of probability densities, which can be applied to any state-space model and which generalize the traditional Kalman filtering methods. Several variants of the particle filter such as SIR, ASIR, and RPF are introduced within a generic framework of the sequential importance sampling (SIS) algorithm. These are discussed and compared with the standard EKF through an illustrative example.

/pdf/a-tutorial-on-particle-filters-for-online-nonlinear-non-1rzkwgk2pp.pdf

A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking

Any series of observations ordered along a single dimension, such as time, may be thought of as a time series. The emphasis in time series analysis is on studying the dependence among observations at different points in time. What distinguishes time series analysis from general multivariate analysis is precisely the temporal order imposed on the observations. Many economic variables, such as GNP and its components, price indices, sales, and stock returns are observed over time. In addition to being interested in the contemporaneous relationships among such variables, we are often concerned with relationships between their current and past values, that is, relationships over time.

Time Series Analysis

计量经济分析 = Econometric analysis

Monte Carlo methods are revolutionizing the on-line analysis of data in fields as diverse as financial modeling, target tracking and computer vision. These methods, appearing under the names of bootstrap filters, condensation, optimal Monte Carlo filters, particle filters and survival of the fittest, have made it possible to solve numerically many complex, non-standard problems that were previously intractable. This book presents the first comprehensive treatment of these techniques, including convergence results and applications to tracking, guidance, automated target recognition, aircraft navigation, robot navigation, econometrics, financial modeling, neural networks, optimal control, optimal filtering, communications, reinforcement learning, signal enhancement, model averaging and selection, computer vision, semiconductor design, population biology, dynamic Bayesian networks, and time series analysis. This will be of great value to students, researchers and practitioners, who have some basic knowledge of probability. Arnaud Doucet received the Ph. D. degree from the University of Paris-XI Orsay in 1997. From 1998 to 2000, he conducted research at the Signal Processing Group of Cambridge University, UK. He is currently an assistant professor at the Department of Electrical Engineering of Melbourne University, Australia. His research interests include Bayesian statistics, dynamic models and Monte Carlo methods. Nando de Freitas obtained a Ph.D. degree in information engineering from Cambridge University in 1999. He is presently a research associate with the artificial intelligence group of the University of California at Berkeley. His main research interests are in Bayesian statistics and the application of on-line and batch Monte Carlo methods to machine learning. Neil Gordon obtained a Ph.D. in Statistics from Imperial College, University of London in 1993. He is with the Pattern and Information Processing group at the Defence Evaluation and Research Agency in the United Kingdom. His research interests are in time series, statistical data analysis, and pattern recognition with a particular emphasis on target tracking and missile guidance.

/pdf/sequential-monte-carlo-methods-in-practice-3kfxeebn8e.pdf

Sequential Monte Carlo methods in practice

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 literatures 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.

/pdf/on-sequential-monte-carlo-sampling-methods-for-bayesian-3bit26usft.pdf

On sequential Monte Carlo sampling methods for Bayesian filtering

A stochastic volatility model may be estimated by a quasi-maximum likelihood procedure by transforming to a linear state-space form. The method is extended to handle correlation between the two disturbances in the model and applied to data on stock returns.

Estimation of an asymmetric model of asset prices

I discuss models which allow the local level model, which rationalised exponentially weighted moving averages, to have a time-varying signal/noise ratio. I call this a martingale component model. This makes the rate of discounting of data local. I show how to handle such models effectively using an auxiliary particle filter which deploys M Kalman filters run in parallel competing against one another. Here one thinks of M as being 1,000 or more. The model is applied to inflation forecasting. The model generalises to unobserved component models where Gaussian shocks are replaced by martingale difference sequences.

Martingale unobserved component models

Likelihood based estimation of the parameters of state space models can be carried out via a particle filter. In this paper we show how to make valid inference on such parameters when the model is incorrect. In particular we develop a simulation strategy for computing sandwich covariance matrices which can be used for asymptotic likelihood based inference. These methods are illustrated on some simulated data.

https://www.nuffield.ox.ac.uk/economics/papers/2012/doucet120601.pdf

Robust inference on parameters via particle filters and sandwich covariance matrices

Econometric analysis of multivariate realised QML: estimation of the covariation of equity prices under asynchronous trading

Importance sampling is used in many aspects of modern econometrics to approximate unsolvable integrals. Its reliable use requires the sampler to possess a variance, for this guarantees a square root speed of convergence and asymptotic normality of the estimator of the integral. However, this assumption is seldom checked. In this paper we propose to use extreme value theory to empirically assess the appropriateness of this assumption. We illustrate this method in the context of a maximum simulated likelihood analysis of the stochastic volatility model.

/pdf/testing-the-assumptions-behind-the-use-of-importance-bjdkljq6rf.pdf

Neil Shephard

Papers

Estimation of an asymmetric model of asset prices

Martingale unobserved component models

Robust inference on parameters via particle filters and sandwich covariance matrices

Econometric analysis of multivariate realised QML: estimation of the covariation of equity prices under asynchronous trading

Testing the assumptions behind the use of importance sampling