CDA Computational Design and Adaptation

Publications

[1]

P. Gaynor, R. Webb, C. Lloyd, and P. Bodger.A low temperature differential stirling engine-based power generation research programme.In Proceedings of the Second IASTED Africa Conference, September 2008.

Research into various forms of sustainable power generation is required to address the rising problems associated with fossil fuel limitations. A new research programme aims to show that electric power generation using low-grade heat and Stirling engine technology can be made commercially viable. Using the Modified Beale Number, and preliminary software modelling, an initial prototype has been designed. This prototype will output approximately 1 kW of electric power with a temperature differential of around 30 K. The fundamental considerations for this prototype are the effects of engine parameters such as compression ratio, pressure, regenerator matrix composition, displacer/power piston phasing, and displacer velocity profile. The result here being a novel gamma-type engine configuration that complements a high pressure system. The prototype also has an electrically actuated displacer piston so that the phasing and velocity profile are finely variable, and a modular regenerator matrix.

[2]

Russell Y. Webb.Multi-layer corrective cascade architecture for on-line predictive echo state networks.Applied Artificial Intelligence, 22(7):811–823, August 2008.(doi:10.1080/08839510802170587)

An architecture for on-line learning of time series prediction is presented which uses a series of echo state networks (ESNs). Each ESN learns to predict an error correction term for the previous ESN. This technique is demonstrated to improve prediction accuracy for on-line learning of the Mackey-Glass chaotic oscillator. The results are compared to other architectural configurations to show that the improved performance emerges from sequential ESN error correction. A new recurrent network structure is shown to be a useful simplification of the usual ESN reservoir.

[3]

Russell Y. Webb.Time series prediction with evolved, composite echo state networks.In Xiaodong, editor, SEAL, volume 5361, pages 555–564. Springer, 2008.

A framework for predictive, on-line, learning networks composed of multiple echo state networks is presented. These composite networks permit learning predictions based on complex combinations of sub-predictions and error terms. The configuration space is explored with a genetic algorithm and better performance is achieved than with hand coded solutions.

[4]

Russell Webb and Peter Smith.A population monte carlo method for generating random matrices with known characteristics.Applied Artificial Intelligence, 22(7):730–748, August 2008.(doi:10.1080/08839510802164143)

An algorithm which provides approximate solutions to a certain matrix inverse problem is presented. In this inverse problem, we usually assume that the distribution of a functional of a random matrix is known. For example, we may know the distribution of the determinant or trace of the matrix. The algorithm attempts to find the mean and covariance structure of a random Gaussian matrix which yields the correct distribution for the functional. The algorithm is based on population Monte Carlo (PMC). Density estimation and importance sampling are used to converge toward a Gaussian matrix solution space described by the means and covariances. We also apply the algorithm to a machine learning problem without a known distribution and show the algorithm can find solutions maximizing an objective function. Results of the algorithm can give insights into the nature of random matrices with certain properties and allow statistical machine learning to create hypotheses about matrix structures from limited measurements. Furthermore, there are applications in testing and communications theory.

[5]

S. J. Weddell and R. Y. Webb.Reservoir computing for prediction of the spatially-variant point spread function.Selected Topics in Signal Processing, IEEE Journal of, 2(5):624–634, 2008.(doi:10.1109/JSTSP.2008.2004218)

A new method is presented which provides prediction of the spatially variant point spread function for the restoration of astronomical images, distorted by atmospheric turbulence when viewed using ground-based telescopes. Our approach uses reservoir computing to firstly learn the spatio-temporal evolution of aberrations caused by turbulence, and secondly, predicts the space-varying point spread function (PSF) for application of widely-used deconvolution algorithms, resulting in the restoration of astronomical images. In this article, a reservoir-based, recurrent neural network is used to predict modal aberrations that comprise the spatially variant PSF over a wide field-of-view using a time-series ensemble from multiple reference beacons.

[6]

S. J. Weddell and R. Y. Webb.The restoration of extended astronomical images using the spatially-variant point spread function.In K. Irie, editor, in proceedings of Image and Vision Computing New Zealand, 2008 (IVCNZ'08), pages 1–6. IEEE Press, Nov 2008.

In this paper we address the problem of image restoration of extended astronomical objects, where the effects of inhomogeneous random media, such as atmospheric turbulence, result in distortion over the pupil plane. Distortions are usually expressed in terms of the point spread function (PSF) and can be used to restore an image with the application of a deconvolution algorithm. Since the PSF is generally spatially dependent, restoration of extended images within specific homogeneous (isoplanatic) regions is required. Inconsistencies at the borders of each region result in error boundaries between restored isoplanatic regions and the object. To minimise such distortion, we determine the optimal size of each isoplanatic region and then apply a linear filter to estimate the PSF between boundaries.

[7]

S.J. Weddell and R.Y. Webb.Reservoir computing for prediction of the spatially-variant point spread function.Selected Topics in Signal Processing, IEEE Journal of, 2(5):624–634, Oct. 2008.(doi:10.1109/JSTSP.2008.2004218)

A new method is presented which provides prediction of the spatially variant point spread function for the restoration of astronomical images, distorted by atmospheric turbulence when viewed using ground-based telescopes. Our approach uses reservoir computing to firstly learn the spatio-temporal evolution of aberrations caused by turbulence, and secondly, predicts the space-varying point spread function (PSF) for application of widely-used deconvolution algorithms, resulting in the restoration of astronomical images. In this article, a reservoir-based, recurrent neural network is used to predict modal aberrations that comprise the spatially variant PSF over a wide field-of-view using a time-series ensemble from multiple reference beacons.

[8]

Russell Y. Webb and Noel C. Macdonald.On-chip electrometer using mems parallel-plate pull-in sensing.International Journal of Intelligent Systems Technologies and Applications, pages 149–162, June 2007.(doi:10.1504/IJISTA.2007.014132)

Microelectromechanical Systems (MEMS) have been proposed as DC electrical metrology references. The design reported here is the first to enhance the qualities of a MEMS DC reference with potential tuning and sensing via an isolated and monolithically integrated MEMS technology and thereby, convert a stable parallel-plate voltage reference to a simple, sensitive, low-burden voltage sensor. This on-chip system reliably measures unknown potentials ranging from -60 to 60V with sampling times less than 10 ms. In the initial design, the system is used to measure atto-amp leakage current though 10 POhm, suspended, MEMS isolation.

[9]

S. J. Weddell and R. Y. Webb.A neural network architecture for the reconstruction of turbulence degraded point spread functions.In M.J. Cree, editor, in proceedings of Image & Vision Computing NZ 2007 (IVCNZ'07), pages 103–108, Hamilton, Dec 2007.

Astronomical images, degraded by the effects of atmospheric turbulence, can be corrected in real-time using adaptive optics (AO). A reconstructed wavefront, measured using a reference object with a small angular separation from a target object, is used to compensate aberrations effecting a target object. The conjugate of the recovered wavefront is used to alter the optical path of a telescope for correction. To ensure high performance, a constraint is imposed on the angular separation between a reference object, such as a bright guide star, and target object which is typically fainter. This constraint is referred to as the isoplanatic angle, θ0. Given the sparsity of natural guide stars, combined with limitations on the use of artificial guide stars, severe restrictions are imposed on the field of view (FOV) and turbulence compensation over the anisoplanatic region. Our aim is to use aberration data from two or more reference objects to extend the isoplantatic angle. This paper outlines a system architecture and proposes the use of artificial neural networks (ANN) to provide extended FOV coverage for real-time astronomical image restoration.

[10]

S. .J. Weddell and R. Y. Webb.Dynamic artificial neural networks for centroid prediction in astronomy.In IEEE Press, editor, Sixth International Conference, Hybrid Intelligent Systems (HIS'06), Auckland, New Zealand, pages 68–72, Auckland, Dec 2006. IEEE Press.

Motivation for this research is the real-time restoration of faint astronomical images through turbulence over a large field-of-view. A simulation platform was developed to predict the centroid of a science object, convolved through multiple perturbation fields, and projected on to an image plane. Centroid data were selected from various source and target locations and used to train an artificial neural network to estimate centroids over a spatial grid, defined on the image plane. The capability of the network to learn to predict centroids over new target locations was assessed using a priori centroid data corresponding to selected grid locations. Various distortion fields were used in training and simulating the network including data collected from observation runs at a local observatory. Results from this work provide the basis for extensions and application to modal tomography.

[11]

Russell Y. Webb and Noel C. Macdonald.Voltage sensing using mems parallel-plate actuation.In International Conference on Sensor Technology, pages 106–111, November 2005.

[12]

S. J. Weddell and R. Y. Webb.Data preprocessing on sequential data for improved astronomical imaging.In B. McCane, editor, in proceedings of Image and Vision Computing New Zealand 2005 (IVCNZ'05), Dunedin, Dec 2005.

Obtaining diffraction limited astronomical images using ground-based telescopes and sensors operating at optical wavelengths is currently not possible due to the effects of atmospheric turbulence. Adaptive optics is a mature technology that attempts to minimise the effects of atmospheric turbulence through wavefront detection and optical path correction in real-time. Our work involves the acquisition and preprocessing of sequential astronomical data and the application of machine learning algorithms to obtain consistently, high-resolution images called lucky frames. This paper will outline the design of a framework used to capture and preprocess sequential data in real-time, and explore machine learning extensions to accurately predict lucky frames.

[13]

C. S. B. Lee, R. Y. Webb, J. M. Chong, and N. C. Macdonald.Single crystal silicon (scs) micromirror arrays using deep silicon etching and ir alignment.In Micro Electro Mechanical Systems, 2000. MEMS 2000. The Thirteenth Annual International Conference on, pages 441–448, 2000.(doi:10.1109/MEMSYS.2000.838558)

10×10 and 5×5, a high mirror fill factor (>70%), low voltage operation (<30 V), single crystal silicon (SCS) micromirror arrays have been designed, fabricated, and tested. Each mirror is 320 μm by 170 μm and is individually addressable. In comparison to thin film micro-mirror arrays, the SCS mirror surface is optically flat and smooth, free of residual stress, and highly reflective after the deposition of a thin aluminum layer. In addition to a flat mirror, high-aspect-ratio grating structures have been fabricated on the surface of the mirrors, enhancing the optical manipulation potential of devices.

[14]

Russell Y. Webb, Scott G. Adams, and Noel C. Macdonald.Suspended thermal oxide trench isolation for scs mems.Microrobotics and Micromanipulation, 3519(1):196–199, 1998.