Category Archives: Bayesian Filtering

Bayesian estimation when computing the likelihood is hard

Kirthevasan Kandasamy, Jeff Schneider, Barnabás Póczos, Query efficient posterior estimation in scientific experiments via Bayesian active learning, Artificial Intelligence, Volume 243, February 2017, Pages 45-56, ISSN 0004-3702, DOI: 10.1016/j.artint.2016.11.002.

A common problem in disciplines of applied Statistics research such as Astrostatistics is of estimating the posterior distribution of relevant parameters. Typically, the likelihoods for such models are computed via expensive experiments such as cosmological simulations of the universe. An urgent challenge in these research domains is to develop methods that can estimate the posterior with few likelihood evaluations.In this paper, we study active posterior estimation in a Bayesian setting when the likelihood is expensive to evaluate. Existing techniques for posterior estimation are based on generating samples representative of the posterior. Such methods do not consider efficiency in terms of likelihood evaluations. In order to be query efficient we treat posterior estimation in an active regression framework. We propose two myopic query strategies to choose where to evaluate the likelihood and implement them using Gaussian processes. Via experiments on a series of synthetic and real examples we demonstrate that our approach is significantly more query efficient than existing techniques and other heuristics for posterior estimation.

Partially observed boolean dynamic systems

M. Imani and U. M. Braga-Neto, “Maximum-Likelihood Adaptive Filter for Partially Observed Boolean Dynamical Systems,” in IEEE Transactions on Signal Processing, vol. 65, no. 2, pp. 359-371, Jan.15, 15 2017.DOI: 10.1109/TSP.2016.2614798.

We present a framework for the simultaneous estimation of state and parameters of partially observed Boolean dynamical systems (POBDS). Simultaneous state and parameter estimation is achieved through the combined use of the Boolean Kalman filter and Boolean Kalman smoother, which provide the minimum mean-square error state estimators for the POBDS model, and maximum-likelihood (ML) parameter estimation; in the presence of continuous parameters, ML estimation is performed using the expectation-maximization algorithm. The performance of the proposed ML adaptive filter is demonstrated by numerical experiments with a POBDS model of gene regulatory networks observed through noisy next-generation sequencing (RNA-seq) time series data using the well-known p53-MDM2 negative-feedback loop gene regulatory model.

A variant of particle filters that uses feedback to model how particles move towards the real posterior

T. Yang, P.~G. Mehta, S.~P. Meyn, Feedback particle filter, IEEE Transactions on Automatic Control, 58 (10) (2013), pp. 2465â–2480, DOI: 10.1109/TAC.2013.2258825.

The feedback particle filter introduced in this paper is a new approach to approximate nonlinear filtering, motivated by techniques from mean-field game theory. The filter is defined by an ensemble of controlled stochastic systems (the particles). Each particle evolves under feedback control based on its own state, and features of the empirical distribution of the ensemble. The feedback control law is obtained as the solution to an optimal control problem, in which the optimization criterion is the Kullback-Leibler divergence between the actual posterior, and the common posterior of any particle. The following conclusions are obtained for diffusions with continuous observations: 1) The optimal control solution is exact: The two posteriors match exactly, provided they are initialized with identical priors. 2) The optimal filter admits an innovation error-based gain feedback structure. 3) The optimal feedback gain is obtained via a solution of an Euler-Lagrange boundary value problem; the feedback gain equals the Kalman gain in the linear Gaussian case. Numerical algorithms are introduced and implemented in two general examples, and a neuroscience application involving coupled oscillators. In some cases it is found that the filter exhibits significantly lower variance when compared to the bootstrap particle filter.

A gentle introduction to Box-Particle Filters

A. Gning, B. Ristic, L. Mihaylova and F. Abdallah, An Introduction to Box Particle Filtering [Lecture Notes], in IEEE Signal Processing Magazine, vol. 30, no. 4, pp. 166-171, July 2013. DOI: 10.1109/MSP.2013.225460.

Resulting from the synergy between the sequential Monte Carlo (SMC) method [1] and interval analysis [2], box particle filtering is an approach that has recently emerged [3] and is aimed at solving a general class of nonlinear filtering problems. This approach is particularly appealing in practical situations involving imprecise stochastic measurements that result in very broad posterior densities. It relies on the concept of a box particle that occupies a small and controllable rectangular region having a nonzero volume in the state space. Key advantages of the box particle filter (box-PF) against the standard particle filter (PF) are its reduced computational complexity and its suitability for distributed filtering. Indeed, in some applications where the sampling importance resampling (SIR) PF may require thousands of particles to achieve accurate and reliable performance, the box-PF can reach the same level of accuracy with just a few dozen box particles. Recent developments [4] also show that a box-PF can be interpreted as a Bayes? filter approximation allowing the application of box-PF to challenging target tracking problems [5].

Using the Bingham distribution of probability, which is defined on a d-dimensional sphere to be antipodally symmetric, to address the problem of angle periodicity in [0,2pi] when estimating orientation in a recursive filter

Gilitschenski, I.; Kurz, G.; Julier, S.J.; Hanebeck, U.D., Unscented Orientation Estimation Based on the Bingham Distribution, in Automatic Control, IEEE Transactions on , vol.61, no.1, pp.172-177, Jan. 2016, DOI: 10.1109/TAC.2015.2423831.

In this work, we develop a recursive filter to estimate orientation in 3D, represented by quaternions, using directional distributions. Many closed-form orientation estimation algorithms are based on traditional nonlinear filtering techniques, such as the extended Kalman filter (EKF) or the unscented Kalman filter (UKF). These approaches assume the uncertainties in the system state and measurements to be Gaussian-distributed. However, Gaussians cannot account for the periodic nature of the manifold of orientations and thus small angular errors have to be assumed and ad hoc fixes must be used. In this work, we develop computationally efficient recursive estimators that use the Bingham distribution. This distribution is defined on the hypersphere and is inherently more suitable for periodic problems. As a result, these algorithms are able to consistently estimate orientation even in the presence of large angular errors. Furthermore, handling of nontrivial system functions is performed using an entirely deterministic method which avoids any random sampling. A scheme reminiscent of the UKF is proposed for the nonlinear manifold of orientations. It is the first deterministic sampling scheme that truly reflects the nonlinear manifold of orientations.

A clarification and systematization of UKF

Menegaz, H.M.T.; Ishihara, J.Y.; Borges, G.A.; Vargas, A.N., A Systematization of the Unscented Kalman Filter Theory, in Automatic Control, IEEE Transactions on , vol.60, no.10, pp.2583-2598, Oct. 2015 DOI: 10.1109/TAC.2015.2404511.

In this paper, we propose a systematization of the (discrete-time) Unscented Kalman Filter (UKF) theory. We gather all available UKF variants in the literature, present corrections to theoretical inconsistencies, and provide a tool for the construction of new UKF’s in a consistent way. This systematization is done, mainly, by revisiting the concepts of Sigma-Representation, Unscented Transformation (UT), Scaled Unscented Transformation (SUT), UKF, and Square-Root Unscented Kalman Filter (SRUKF). Inconsistencies are related to 1) matching the order of the transformed covariance and cross-covariance matrices of both the UT and the SUT; 2) multiple UKF definitions; 3) issue with some reduced sets of sigma points described in the literature; 4) the conservativeness of the SUT; 5) the scaling effect of the SUT on both its transformed covariance and cross-covariance matrices; and 6) possibly ill-conditioned results in SRUKF’s. With the proposed systematization, the symmetric sets of sigma points in the literature are formally justified, and we are able to provide new consistent variations for UKF’s, such as the Scaled SRUKF’s and the UKF’s composed by the minimum number of sigma points. Furthermore, our proposed SRUKF has improved computational properties when compared to state-of-the-art methods.

Modelling ECGs with sums of gaussians and estimating them through switching Kalman Filters and the likelihood of each mode

Oster, J.; Behar, J.; Sayadi, O.; Nemati, S.; Johnson, A.E.W.; Clifford, G.D., Semisupervised ECG Ventricular Beat Classification With Novelty Detection Based on Switching Kalman Filters, in Biomedical Engineering, IEEE Transactions on , vol.62, no.9, pp.2125-2134, Sept. 2015, DOI: 10.1109/TBME.2015.2402236.

Automatic processing and accurate diagnosis of pathological electrocardiogram (ECG) signals remains a challenge. As long-term ECG recordings continue to increase in prevalence, driven partly by the ease of remote monitoring technology usage, the need to automate ECG analysis continues to grow. In previous studies, a model-based ECG filtering approach to ECG data from healthy subjects has been applied to facilitate accurate online filtering and analysis of physiological signals. We propose an extension of this approach, which models not only normal and ventricular heartbeats, but also morphologies not previously encountered. A switching Kalman filter approach is introduced to enable the automatic selection of the most likely mode (beat type), while simultaneously filtering the signal using appropriate prior knowledge. Novelty detection is also made possible by incorporating a third mode for the detection of unknown (not previously observed) morphologies, and denoted as X-factor. This new approach is compared to state-of-the-art techniques for the ventricular heartbeat classification in the MIT-BIH arrhythmia and Incart databases. F1 scores of 98.3% and 99.5% were found on each database, respectively, which are superior to other published algorithms’ results reported on the same databases. Only 3% of all the beats were discarded as X-factor, and the majority of these beats contained high levels of noise. The proposed technique demonstrates accurate beat classification in the presence of previously unseen (and unlearned) morphologies and noise, and provides an automated method for morphological analysis of arbitrary (unknown) ECG leads.

A novel non-linear bayesian filter for continuous time estimation with a nice comparison to discrete-time filters

Atiyeh Ghoreyshi and Terence D. Sanger, A Nonlinear Stochastic Filter for Continuous-Time State Estimation, IEEE Transactions on Automatic Control, vol. 60, no. 8, DOI: 10.1109/TAC.2015.2409910.

Nonlinear filters produce a nonparametric estimate of the probability density of state at each point in time. Currently known nonlinear filters include Particle Filters and the Kushner equation (and its un-normalized version: the Zakai equation). However, these filters have limited measurement models: Particle Filters require measurement at discrete times, and the Kushner and Zakai equations only apply when the measurement can be represented as a function of the state. We present a new nonlinear filter for continuous-time measurements with a much more general stochastic measurement model. It integrates to Bayes’ rule over short time intervals and provides Bayes-optimal estimates from quantized, intermittent, or ambiguous sensor measurements. The filter has a close link to Information Theory, and we show that the rate of change of entropy of the density estimate is equal to the mutual information between the measurement and the state and thus the maximum achievable. This is a fundamentally new class of filter that is widely applicable to nonlinear estimation for continuous-time control.

Substituting the update step of a bayesian filter by a maximum likelihood optimisation in order to use non-linear observation models in a (linear-transition) Kalman framework

Damián Marelli, Minyue Fu, and Brett Ninness, Asymptotic Optimality of the Maximum-Likelihood Kalman Filter for Bayesian Tracking With Multiple Nonlinear Sensors, IEEE Transactions on signal processing, vol. 63, no. 17, DOI: 10.1109/TSP.2015.2440220.

Bayesian tracking is a general technique for state estimation of nonlinear dynamic systems, but it suffers from the drawback of computational complexity. This paper is concerned with a class of Wiener systems with multiple nonlinear sensors. Such a system consists of a linear dynamic system followed by a set of static nonlinear measurements. We study a maximum-likelihood Kalman filtering (MLKF) technique which involves maximum-like-lihood estimation of the nonlinear measurements followed by classical Kalman filtering. This technique permits a distributed implementation of the Bayesian tracker and guarantees the boundedness of the estimation error. The focus of this paper is to study the extent to which the MLKF technique approximates the theoretically optimal Bayesian tracker. We provide conditions to guarantee that this approximation becomes asymptotically exact as the number of sensors becomes large. Two case studies are analyzed in detail.

A brief general explanation of Rao-Blacwellization and a new way of applying it to reduce the variance of a point estimation in a sequential bayesian setting

Petetin, Y.; Desbouvries, F., Bayesian Conditional Monte Carlo Algorithms for Nonlinear Time-Series State Estimation, Signal Processing, IEEE Transactions on , vol.63, no.14, pp.3586,3598, DOI: 10.1109/TSP.2015.2423251.

Bayesian filtering aims at estimating sequentially a hidden process from an observed one. In particular, sequential Monte Carlo (SMC) techniques propagate in time weighted trajectories which represent the posterior probability density function (pdf) of the hidden process given the available observations. On the other hand, conditional Monte Carlo (CMC) is a variance reduction technique which replaces the estimator of a moment of interest by its conditional expectation given another variable. In this paper, we show that up to some adaptations, one can make use of the time recursive nature of SMC algorithms in order to propose natural temporal CMC estimators of some point estimates of the hidden process, which outperform the associated crude Monte Carlo (MC) estimator whatever the number of samples. We next show that our Bayesian CMC estimators can be computed exactly, or approximated efficiently, in some hidden Markov chain (HMC) models; in some jump Markov state-space systems (JMSS); as well as in multitarget filtering. Finally our algorithms are validated via simulations.