Competitive training in hidden Markov models

The use of hidden Markov models is placed in a connectionist framework, and an alternative approach to improving their ability to discriminate between classes is described. Using a network style of training, a measure of discrimination based on the a posteriori probability of state occupation is proposed, and the theory for its optimization using error back-propagation and gradient ascent is presented. The method is shown to be numerically well behaved, and results are presented which demonstrate that when using a simple threshold test on the probability of state occupation, the proposed optimization scheme leads to improved recognition performance.

Beam propagation modelling of tapered waveguide lasers

A dynamic beam propagation model allows design optimization of high power low divergence tapered waveguide lasers. The model is extended to include spatially-resolved temperature profiles and a temperature dependent gain. Using this model, design parameters such as the optimum facet reflectivity, taper angle, and waveguide dimension can be calculated for low far-field divergence and high continuous wave power.

The modified Kanerva model for automatic speech recognition

A parallel processing network derived from Kanerva's associative memory theory Kanerva 1984 is shown to be able to train rapidly on connected speech data and recognize further speech data with a label error rate of 0·68%. This modified Kanerva model can be trained substantially faster than other networks with comparable pattern discrimination properties. Kanerva presented his theory of a self-propagating search in 1984, and showed theoretically that large-scale versions of his model would have powerful pattern matching properties. This paper describes how the design for the modified Kanerva model is derived from Kanerva's original theory. Several designs are tested to discover which form may be implemented fastest while still maintaining versatile recognition performance. A method is developed to deal with the time varying nature of the speech signal by recognizing static patterns together with a fixed quantity of contextual information. In order to recognize speech features in different contexts it is necessary for a network to be able to model disjoint pattern classes. This type of modelling cannot be performed by a single layer of links. Network research was once held back by the inability of single-layer networks to solve this sort of problem, and the lack of a training algorithm for multi-layer networks. Rumelhart, Hinton & Williams 1985 provided one solution by demonstrating the "back propagation" training algorithm for multi-layer networks. A second alternative is used in the modified Kanerva model. A non-linear fixed transformation maps the pattern space into a space of higher dimensionality in which the speech features are linearly separable. A single-layer network may then be used to perform the recognition. The advantage of this solution over the other using multi-layer networks lies in the greater power and speed of the single-layer network training algorithm. © 1989.

Contrasting code performances for computational aeroacoustics of jets

The numerical propagation of subcritical Tollmein-Schlichting (T-S), inviscid vortical and cut-on acoustic waves is explored. For the former case, the performances of the very different NEAT, NTS, HYDRA, FLUXp and OSMIS3D codes is studied. A modest/coarse hexahedral computational grid that starkly shows differences between the different codes and schemes used in them is employed. For the same order of discretization the five codes show similar results. The unstructured codes are found to propagate vortical and acoustic waves well on triangular cell meshes but not the T-S wave. The above code contrasting exercise is then carried out using implicit LES or Smagorinsky LES for and Ma = 0.9 plane jet on modest 0.5 million cell grids moving to circa 5 million cell grids. For this case, even on the coarse grid, for all codes results were generally encouraging. In general, the spread in computational results is less than the spread of the measurements. Interestingly, the finer grid turbulence intensity levels are slightly more under-predicted than those of the coarse grid. This difference is attributed to the numerical dispersion error having a favourable coarse grid influence. For a non-isothermal jet, HYDRA and NTS also give encouraging results. Peak turbulence values along the jet centreline are in better agreement with measurements than for the isothermal jets. Copyright © 2006 by University of Wales.

X-ray propagation in carbon nanotubes

The maintenance of the growth of the multibillion-dollar semiconductor industry requires the development of techniques for the fabrication and characterisation of nanoscale devices. Consequently, there is great interest in photolithography techniques such as extreme UV and x-ray. Both of these techniques are extremely expensive and technologically very demanding. In this paper we describe research on the feasibility of exploiting x-ray propagation within carbon nanotubes (CNT's) for the fabrication and characterisation of nanoscale devices. This work discusses the parameters determining the design space available. To demonstrate experimentally the feasibility of x-ray propagation, arrays of carbon nanotubes have been grown on silicon membranes. The latter are required to provide structural support for the CNT's while minimising energy loss. To form a waveguide metal is deposited between the nanotubes to block x-ray transmission in this region at the same time as cladding the CNT's. The major challenge has been to fill the spaces between the CNT's with material of sufficient thickness to block x-ray transmission while maintaining the structural integrity of the CNT's. Various techniques have been employed to fill the gaps between the nanotubes including electroplating, sputtering and evaporation. This work highlights challenges encountered in optimising the process.

Once again on the importance of propagation effects for jet noise modelling

Several turbulent jet noise models starting from the classical Lighthill acoustic analogy to state-of-the art models are considered. No attempt is made to present any complete overview of jet noise theories. Instead, the aim is to emphasise the importance of sound generation and meanflow effects for the understanding and prediction of jet noise. For a recent acoustic analogy model, the consequences of jet flow simplification on the predicted sound spectra shape and the effective noise source location in the jet are discussed. © 2010 by the American Institute of Aeronautics and Astronautics, Inc.

CFD-DEM simulation of propagation of sound waves in fluid particles fluidised medium

In this work, speed of sound in 2 phase mixture has been explored using CFD-DEM (Computational Fluid Dynamcis - Discrete Element Modelling). In this method volume averaged Navier Stokes, continuity and energy equations are solved for fluid. Particles are simulated as individual entities; their behaviour is captured by Newton's laws of motion and classical contact mechanics. Particle-fluid interaction is captured using drag laws given in literature.The speed of sound in a medium depends on physical properties. It has been found experimentally that speed of sound drops significantly in 2 phase mixture of fluidised particles because of its increased density relative to gas while maintaining its compressibility. Due to the high rate of heat transfer within 2 phase medium as given in Roy et al. (1990), it has been assumed that the fluidised gas-particle medium is isothermal.The similar phenomenon has been tried to be captured using CFD-DEM numerical simulation. The disturbance is introduced and fundamental frequency in the medium is noted to measure the speed of sound for e.g. organ pipe. It has been found that speed of sound is in agreement with the relationship given in Roy et al. (1990). Their assumption that the system is isothermal also appears to be valid.

The effect of a twin tunnel on the propagation of ground-borne vibration from an underground railway

Accurate predictions of ground-borne vibration levels in the vicinity of an underground railway are greatly sought after in modern urban centres. Yet the complexity involved in simulating the underground environment means that it is necessary to make simplifying assumptions about this system. One such commonly made assumption is to ignore the effects of neighbouring tunnels, despite the fact that many underground railway lines consist of twin-bored tunnels, one for the outbound direction and one for the inbound direction. This paper presents a unique model for two tunnels embedded in a homogeneous, elastic fullspace. Each of these tunnels is subject to both known, dynamic train forces and dynamic cavity forces. The net forces acting on the tunnels are written as the sum of those tractions acting on the invert of a single tunnel, and those tractions that represent the motion induced by the neighbouring tunnel. By apportioning the tractions in this way, the vibration response of a two-tunnel system is written as a linear combination of displacement fields produced by a single-tunnel system. Using Fourier decomposition, forces are partitioned into symmetric and antisymmetric modenumber components to minimise computation times. The significance of the interactions between two tunnels is quantified by calculating the insertion gains, in both the vertical and horizontal directions, that result from the existence of a second tunnel. The insertion-gain results are shown to be localised and highly dependent on frequency, tunnel orientation and tunnel thickness. At some locations, the magnitude of these insertion gains is greater than 20 dB. This demonstrates that a high degree of inaccuracy exists in any surface vibration prediction model that includes only one of the two tunnels. This novel two-tunnel solution represents a significant contribution to the existing body of research into vibration from underground railways, as it shows that the second tunnel has a significant influence on the accuracy of vibration predictions for underground railways. © 2011 Elsevier Ltd. All rights reserved.