State estimation schemes for independent component coupled hidden Markov models

Conventional Hidden Markov models generally consist of a Markov chain observed through a linear map corrupted by additive noise. This general class of model has enjoyed a huge and diverse range of applications, for example, speech processing, biomedical signal processing and more recently quantitative finance. However, a lesser known extension of this general class of model is the so-called Factorial Hidden Markov Model (FHMM). FHMMs also have diverse applications, notably in machine learning, artificial intelligence and speech recognition [13, 17]. FHMMs extend the usual class of HMMs, by supposing the partially observed state process is a finite collection of distinct Markov chains, either statistically independent or dependent. There is also considerable current activity in applying collections of partially observed Markov chains to complex action recognition problems, see, for example, [6]. In this article we consider the Maximum Likelihood (ML) parameter estimation problem for FHMMs. Much of the extant literature concerning this problem presents parameter estimation schemes based on full data log-likelihood EM algorithms. This approach can be slow to converge and often imposes heavy demands on computer memory. The latter point is particularly relevant for the class of FHMMs where state space dimensions are relatively large. The contribution in this article is to develop new recursive formulae for a filter-based EM algorithm that can be implemented online. Our new formulae are equivalent ML estimators, however, these formulae are purely recursive and so, significantly reduce numerical complexity and memory requirements. A computer simulation is included to demonstrate the performance of our results. © Taylor & Francis Group, LLC.

A reversible infinite HMM using normalised random measures

Copyright 2014 by the author(s). We present a nonparametric prior over reversible Markov chains. We use completely random measures, specifically gamma processes, to construct a countably infinite graph with weighted edges. By enforcing symmetry to make the edges undirected we define a prior over random walks on graphs that results in a reversible Markov chain. The resulting prior over infinite transition matrices is closely related to the hierarchical Dirichlet process but enforces reversibility. A reinforcement scheme has recently been proposed with similar properties, but the de Finetti measure is not well characterised. We take the alternative approach of explicitly constructing the mixing measure, which allows more straightforward and efficient inference at the cost of no longer having a closed form predictive distribution. We use our process to construct a reversible infinite HMM which we apply to two real datasets, one from epigenomics and one ion channel recording.

Multiple object tracking using evolutionary MCMC-based particle algorithms

Algorithms are presented for detection and tracking of multiple clusters of co-ordinated targets. Based on a Markov chain Monte Carlo sampling mechanization, the new algorithms maintain a discrete approximation of the filtering density of the clusters' state. The filters' tracking efficiency is enhanced by incorporating various sampling improvement strategies into the basic Metropolis-Hastings scheme. Thus, an evolutionary stage consisting of two primary steps is introduced: 1) producing a population of different chain realizations, and 2) exchanging genetic material between samples in this population. The performance of the resulting evolutionary filtering algorithms is demonstrated in two different settings. In the first, both group and target properties are estimated whereas in the second, which consists of a very large number of targets, only the clustering structure is maintained. © 2009 IFAC.

A Newton algorithm for invariant subspace computation with large basins of attraction

We study the global behaviour of a Newton algorithm on the Grassmann manifold for invariant subspace computation. It is shown that the basins of attraction of the invariant subspaces may collapse in case of small eigenvalue gaps. A Levenberg-Marquardt-like modification of the algorithm with low numerical cost is proposed. A simple strategy for choosing the parameter is shown to dramatically enlarge the basins of attraction of the invariant subspaces while preserving the fast local convergence.

Bayesian model selection for time series using Markov chain Monte Carlo

We present a stochastic simulation technique for subset selection in time series models, based on the use of indicator variables with the Gibbs sampler within a hierarchical Bayesian framework. As an example, the method is applied to the selection of subset linear AR models, in which only significant lags are included. Joint sampling of the indicators and parameters is found to speed convergence. We discuss the possibility of model mixing where the model is not well determined by the data, and the extension of the approach to include non-linear model terms.

Bayesian learning of noisy Markov decision processes

This work addresses the problem of estimating the optimal value function in a Markov Decision Process from observed state-action pairs. We adopt a Bayesian approach to inference, which allows both the model to be estimated and predictions about actions to be made in a unified framework, providing a principled approach to mimicry of a controller on the basis of observed data. A new Markov chain Monte Carlo (MCMC) sampler is devised for simulation from theposterior distribution over the optimal value function. This step includes a parameter expansion step, which is shown to be essential for good convergence properties of the MCMC sampler. As an illustration, the method is applied to learning a human controller.

Large port count high-speed optical switch fabric for use within datacenters [Invited]

This paper reviews advances in the technology of integrated semiconductor optical amplifier based photonic switch fabrics, with particular emphasis on their suitability for high performance network switches for use within a datacenter. The key requirements for large port count optical switch fabrics are addressed noting the need for switches with substantial port counts. The design options for a 16×16 port photonic switch fabric architecture are discussed and the choice of a Clos-tree design is described. The control strategy, based on arbitration and scheduling, for an integrated switch fabric is explained. The detailed design and fabrication of the switch is followed by experimental characterization, showing net optical gain and operation at 10 Gb/s with bit error rates lower than 10-9. Finally improvements to the switch are suggested, which should result in 100 Gb/s per port operation at energy efficiencies of 3 pJ/bit. © 2011 Optical Society of America.

The design, development, and manufacture of large wood/composite wind turbine rotors

Discusses a study conducted to determine the best development path for large wind turbine rotor design. Shape and number of blades, degrees of freedom allowed, and control strategy are considered. Manufacture and costs are also discussed. Two-bladed, stall-regulated, teetered rotors are more cost effective than three-bladed rotors. Single-bladed rotors can be even more cost-effective. No new manufacturing techniques are required. The most cost-effective rotor includes a hub constructed in wood/composite materials, bonded to the blades. There is strong incentive for the blade manufacturer to supply the complete rotor. (from author's abstract)

Blind equalization and sequence estimation using Markov chain Monte Carlo methods

This paper discusses the problem of restoring a digital input signal that has been degraded by an unknown FIR filter in noise, using the Gibbs sampler. A method for drawing a random sample of a sequence of bits is presented; this is shown to have faster convergence than a scheme by Chen and Li, which draws bits independently. ©1998 IEEE.

Large-eddy simulation of complex geometry jets

Computations are made for chevron and coflowing jet nozzles. The latter has a bypass ratio of 6:1. Also, unlike the chevron nozzle, the core flow is heated, making the inlet conditions reminiscent of those for a real engine. A large-eddy resolving approach is used with circa 12 × 10 6 cell meshes. Because the codes being used tend toward being dissipative the subgrid scale model is abandoned, giving what can be termed numerical large-eddy simulation. To overcome near-wall modeling problems a hybrid numerical large-eddy simulation-Reynolds-averaged Navier-Stokes related method is used. For y + ≤ 60 a Reynolds-averaged Navier-Stokes model is used. Blending between the two regions makes use of the differential Hamilton-Jabobi equation, an extension of the eikonal equation. For both nozzles, results show encouraging agreement with measurements of other workers. The eikonal equation is also used for ray tracing to explore the effect of the mean flow on acoustic ray trajectories, thus yielding a coherent solution strategy. © 2011 by Cambridge University.

Large eddy simulation of a hot, high speed coflowing jet

Computations are made of a short cowl coflowing jet nozzle with a bypass ratio 8 : 1. The core flow is heated, making the inlet conditions reminiscent of those for a real engine. A large eddy resolving approach is used with a 12 × 106 cell mesh. Since the code being used tends towards being dissipative the sub-grid scale (SGS) model is abandoned giving what can be termed Numerical Large Eddy Simulation (NLES). To overcome near wall modelling problems a hybrid NLES-RANS (Reynolds Averaged Navier-Stokes) related method is used. For y+ ≤ 60 a κ-l model is used. Blending between the two regions makes use of the differential Hamilton-Jabobi (HJ) equation, an extension of the eikonal equation. Results show encouraging agreement with existing measurements of other workers. The eikonal equation is also used for acoustic ray tracing to explore the effect of the mean flow on acoustic ray trajectories, thus yielding a coherent solution strategy. Copyright © 2011 by ASME.