Inducing shades of meaning by matrix methods : a first step towards thematic analysis of opinion

This article explores two matrix methods to induce the ``shades of meaning" (SoM) of a word. A matrix representation of a word is computed from a corpus of traces based on the given word. Non-negative Matrix Factorisation (NMF) and Singular Value Decomposition (SVD) compute a set of vectors corresponding to a potential shade of meaning. The two methods were evaluated based on loss of conditional entropy with respect to two sets of manually tagged data. One set reflects concepts generally appearing in text, and the second set comprises words used for investigations into word sense disambiguation. Results show that for NMF consistently outperforms SVD for inducing both SoM of general concepts as well as word senses. The problem of inducing the shades of meaning of a word is more subtle than that of word sense induction and hence relevant to thematic analysis of opinion where nuances of opinion can arise.

Krylov subspace methods for approximating functions of symmetric positive definite matrices with applications to applied statistics and anomalous diffusion

Matrix function approximation is a current focus of worldwide interest and finds application in a variety of areas of applied mathematics and statistics. In this thesis we focus on the approximation of A^(-α/2)b, where A ∈ ℝ^(n×n) is a large, sparse symmetric positive definite matrix and b ∈ ℝ^n is a vector. In particular, we will focus on matrix function techniques for sampling from Gaussian Markov random fields in applied statistics and the solution of fractional-in-space partial differential equations. Gaussian Markov random fields (GMRFs) are multivariate normal random variables characterised by a sparse precision (inverse covariance) matrix. GMRFs are popular models in computational spatial statistics as the sparse structure can be exploited, typically through the use of the sparse Cholesky decomposition, to construct fast sampling methods. It is well known, however, that for sufficiently large problems, iterative methods for solving linear systems outperform direct methods. Fractional-in-space partial differential equations arise in models of processes undergoing anomalous diffusion. Unfortunately, as the fractional Laplacian is a non-local operator, numerical methods based on the direct discretisation of these equations typically requires the solution of dense linear systems, which is impractical for fine discretisations. In this thesis, novel applications of Krylov subspace approximations to matrix functions for both of these problems are investigated. Matrix functions arise when sampling from a GMRF by noting that the Cholesky decomposition A = LL^T is, essentially, a `square root' of the precision matrix A. Therefore, we can replace the usual sampling method, which forms x = L^(-T)z, with x = A^(-1/2)z, where z is a vector of independent and identically distributed standard normal random variables. Similarly, the matrix transfer technique can be used to build solutions to the fractional Poisson equation of the form ϕn = A^(-α/2)b, where A is the finite difference approximation to the Laplacian. Hence both applications require the approximation of f(A)b, where f(t) = t^(-α/2) and A is sparse. In this thesis we will compare the Lanczos approximation, the shift-and-invert Lanczos approximation, the extended Krylov subspace method, rational approximations and the restarted Lanczos approximation for approximating matrix functions of this form. A number of new and novel results are presented in this thesis. Firstly, we prove the convergence of the matrix transfer technique for the solution of the fractional Poisson equation and we give conditions by which the finite difference discretisation can be replaced by other methods for discretising the Laplacian. We then investigate a number of methods for approximating matrix functions of the form A^(-α/2)b and investigate stopping criteria for these methods. In particular, we derive a new method for restarting the Lanczos approximation to f(A)b. We then apply these techniques to the problem of sampling from a GMRF and construct a full suite of methods for sampling conditioned on linear constraints and approximating the likelihood. Finally, we consider the problem of sampling from a generalised Matern random field, which combines our techniques for solving fractional-in-space partial differential equations with our method for sampling from GMRFs.

Contradictions : Genre, causal layered analysis and analysis of institutional change

In what follows, I put forward an argument for an analytical method for social science that operates at the level of genre. I argue that generic convergence, generic hybridity, and generic instability provide us with a powerful perspectives on changes in political, cultural, and economic relationships, most specifically at the level of institutions. Such a perspective can help us identify the transitional elements, relationships, and trajectories that define the place of our current system in history, thereby grounding our understanding of possible futures.1 In historically contextualising our present with this method, my concern is to indicate possibilities for the future. Systemic contradictions indicate possibility spaces within which systemic change must and will emerge. We live in a system currently dominated by many fully-expressed contradictions, and so in the presence of many possible futures. The contradictions of the current age are expressed most overtly in the public genres of power politics. Contemporary public policy—indeed politics in general-is an excellent focus for any investigation of possible futures, precisely because of its future-oriented function. It is overtly hortatory; it is designed ‘to get people to do things’ (Muntigl in press: 147). There is no point in trying to get people to do things in the past. Consequently, policy discourse is inherently oriented towards creating some future state of affairs (Graham in press), along with concomitant ways of being, knowing, representing, and acting (Fairclough 2000).

Biosecurity threats : the design of surveillance systems, based on power and risk

We consider the problem of designing a surveillance system to detect a broad range of invasive species across a heterogeneous sampling frame. We present a model to detect a range of invertebrate invasives whilst addressing the challenges of multiple data sources, stratifying for differential risk, managing labour costs and providing sufficient power of detection.We determine the number of detection devices required and their allocation across the landscape within limiting resource constraints. The resulting plan will lead to reduced financial and ecological costs and an optimal surveillance system.

Population Monte Carlo Algorithm in High Dimensions

The population Monte Carlo algorithm is an iterative importance sampling scheme for solving static problems. We examine the population Monte Carlo algorithm in a simplified setting, a single step of the general algorithm, and study a fundamental problem that occurs in applying importance sampling to high-dimensional problem. The precision of the computed estimate from the simplified setting is measured by the asymptotic variance of estimate under conditions on the importance function. We demonstrate the exponential growth of the asymptotic variance with the dimension and show that the optimal covariance matrix for the importance function can be estimated in special cases.

Reliability driven reconfiguration of rural power distribution systems

This paper presents a reliability-based reconfiguration methodology for power distribution systems. Probabilistic reliability models of the system components are considered and Monte Carlo method is used while evaluating the reliability of the distribution system. The reconfiguration is aimed at maximizing the reliability of the power supplied to the customers. A binary particle swarm optimization (BPSO) algorithm is used as a tool to determine the optimal configuration of the sectionalizing and tie switches in the system. The proposed methodology is applied on a modified IEEE 13-bus distribution system.

Power sharing and stability enhancement of an autonomous microgrid with inertial and non-inertial DGs with DSTATCOM

This paper proposes a method of enhancing system stability with a distribution static compensator (DSTATCOM) in an autonomous microgrid with multiple distributed generators (DG). It is assumed that there are both inertial and non-inertial DGs connected to the microgrid. The inertial DG can be a synchronous machine of smaller rating while inertia less DGs (solar) are assumed as DC sources. The inertia less DGs are connected through Voltage Source Converter (VSC) to the microgrid. The VSCs are controlled by either state feedback or current feedback mode to achieve desired voltage-current or power outputs respectively. The power sharing among the DGs is achieved by drooping voltage angle. Once the reference for the output voltage magnitude and angle is calculated from the droop, state feedback controllers are used to track the reference. The angle reference for the synchronous machine is compared with the output voltage angle of the machine and the error is fed to a PI controller. The controller output is used to set the power reference of the synchronous machine. The rate of change in the angle in a synchronous machine is restricted by the machine inertia and to mimic this nature, the rate of change in the VSCs angles are restricted by a derivative feedback in the droop control. The connected distribution static compensator (DSTATCOM) provides ride through capability during power imbalance in the microgrid, especially when the stored energy of the inertial DG is not sufficient to maintain stability. The inclusion of the DSATCOM in such cases ensures the system stability. The efficacies of the controllers are established through extensive simulation studies using PSCAD.

Numerical investigations of linear least squares methods for derivative estimation

The results of a numerical investigation into the errors for least squares estimates of function gradients are presented. The underlying algorithm is obtained by constructing a least squares problem using a truncated Taylor expansion. An error bound associated with this method contains in its numerator terms related to the Taylor series remainder, while its denominator contains the smallest singular value of the least squares matrix. Perhaps for this reason the error bounds are often found to be pessimistic by several orders of magnitude. The circumstance under which these poor estimates arise is elucidated and an empirical correction of the theoretical error bounds is conjectured and investigated numerically. This is followed by an indication of how the conjecture is supported by a rigorous argument.

Peripheral ocular monochromatic aberrations

Aberrations affect image quality of the eye away from the line of sight as well as along it. High amounts of lower order aberrations are found in the peripheral visual field and higher order aberrations change away from the centre of the visual field. Peripheral resolution is poorer than that in central vision, but peripheral vision is important for movement and detection tasks (for example driving) which are adversely affected by poor peripheral image quality. Any physiological process or intervention that affects axial image quality will affect peripheral image quality as well. The aim of this study was to investigate the effects of accommodation, myopia, age, and refractive interventions of orthokeratology, laser in situ keratomileusis and intraocular lens implantation on the peripheral aberrations of the eye. This is the first systematic investigation of peripheral aberrations in a variety of subject groups. Peripheral aberrations can be measured either by rotating a measuring instrument relative to the eye or rotating the eye relative to the instrument. I used the latter as it is much easier to do. To rule out effects of eye rotation on peripheral aberrations, I investigated the effects of eye rotation on axial and peripheral cycloplegic refraction using an open field autorefractor. For axial refraction, the subjects fixated at a target straight ahead, while their heads were rotated by ±30º with a compensatory eye rotation to view the target. For peripheral refraction, the subjects rotated their eyes to fixate on targets out to ±34° along the horizontal visual field, followed by measurements in which they rotated their heads such that the eyes stayed in the primary position relative to the head while fixating at the peripheral targets. Oblique viewing did not affect axial or peripheral refraction. Therefore it is not critical, within the range of viewing angles studied, if axial and peripheral refractions are measured with rotation of the eye relative to the instrument or rotation of the instrument relative to the eye. Peripheral aberrations were measured using a commercial Hartmann-Shack aberrometer. A number of hardware and software changes were made. The 1.4 mm range limiting aperture was replaced by a larger aperture (2.5 mm) to ensure all the light from peripheral parts of the pupil reached the instrument detector even when aberrations were high such as those occur in peripheral vision. The power of the super luminescent diode source was increased to improve detection of spots passing through the peripheral pupil. A beam splitter was placed between the subjects and the aberrometer, through which they viewed an array of targets on a wall or projected on a screen in a 6 row x 7 column matrix of points covering a visual field of 42 x 32. In peripheral vision, the pupil of the eye appears elliptical rather than circular; data were analysed off-line using custom software to determine peripheral aberrations. All analyses in the study were conducted for 5.0 mm pupils. Influence of accommodation on peripheral aberrations was investigated in young emmetropic subjects by presenting fixation targets at 25 cm and 3 m (4.0 D and 0.3 D accommodative demands, respectively). Increase in accommodation did not affect the patterns of any aberrations across the field, but there was overall negative shift in spherical aberration across the visual field of 0.10 ± 0.01m. Subsequent studies were conducted with the targets at a 1.2 m distance. Young emmetropes, young myopes and older emmetropes exhibited similar patterns of astigmatism and coma across the visual field. However, the rate of change of coma across the field was higher in young myopes than young emmetropes and was highest in older emmetropes amongst the three groups. Spherical aberration showed an overall decrease in myopes and increase in older emmetropes across the field, as compared to young emmetropes. Orthokeratology, spherical IOL implantation and LASIK altered peripheral higher order aberrations considerably, especially spherical aberration. Spherical IOL implantation resulted in an overall increase in spherical aberration across the field. Orthokeratology and LASIK reversed the direction of change in coma across the field. Orthokeratology corrected peripheral relative hypermetropia through correcting myopia in the central visual field. Theoretical ray tracing demonstrated that changes in aberrations due to orthokeratology and LASIK can be explained by the induced changes in radius of curvature and asphericity of the cornea. This investigation has shown that peripheral aberrations can be measured with reasonable accuracy with eye rotation relative to the instrument. Peripheral aberrations are affected by accommodation, myopia, age, orthokeratology, spherical intraocular lens implantation and laser in situ keratomileusis. These factors affect the magnitudes and patterns of most aberrations considerably (especially coma and spherical aberration) across the studied visual field. The changes in aberrations across the field may influence peripheral detection and motion perception. However, further research is required to investigate how the changes in aberrations influence peripheral detection and motion perception and consequently peripheral vision task performance.

Design and evaluation of an image analysis platform for low-power, low-bandwidth camera networks

We describe the design and evaluation of a platform for networks of cameras in low-bandwidth, low-power sensor networks. In our work to date we have investigated two different DSP hardware/software platforms for undertaking the tasks of compression and object detection and tracking. We compare the relative merits of each of the hardware and software platforms in terms of both performance and energy consumption. Finally we discuss what we believe are the ongoing research questions for image processing in WSNs.