Browsing by Author "Greene, J R"
Now showing 1 - 6 of 6
Results Per Page
Sort Options
- ItemOpen AccessA comparison of three class separability measures(2004) Mthembu, N S; Greene, J RMeasures of class separability can provide valuable insights into data, and suggest promising classification algorithms and approaches in data mining. We compare three simple class separability measures used in supervised machine learning. Their relative effectiveness is evaluated through their functional relationships and their random projections of data onto R 2 for visualization. We conclude that the simple direct class separability measure of a dataset is an easier and more informative measure for separability than the class scatter matrices approach and it correlates well with Thornton’s Separability’s index.
- ItemOpen AccessA simple method for visualizing labelled and unlabelled data in high-dimensional spaces(2004) Greene, J RThe low-dimensional visualisation of highdimensional data is a valuable way of detecting structure (such as clusters, and the presence of outliers) in the data, and avoiding some of the pitfalls of blind data manipulation. Projection based on principal component analysis is widely employed and often useful, but it is a variancepreserving projection which takes no account of class labels, and may, for this reason, hide significant structure. Here we present a very simple method which appears to yield useful visualizations for many datasets. It is based on a random search for a linear transformation, and projection into a twodimensional visual space, which maximises an objective measure of class separability in the visual space. The method, which can be thought of as a variant of projection pursuit with a novel interest measure, is demonstrated on datasets from the UCI Repository. Tentative interim results are also given for a proposed extension based on spectral clustering, for extending the method to unlabelled data.
- ItemOpen AccessInvestigation into the feasibility of an electro-optic rail surface monitor(1993) Rawstorne, Alexander F; Greene, J RCracks and surface imperfections in railway track are important when considering its lifetime. This thesis investigates the feasibility of a new electro-optic technique which records surface irregularities on a track. It is more sensitive to crack detection than the traditional trailing wheel method and approximately one thousand times less expensive than the highly sensitive ultrasonic technique used by British Rail. The electro-optic method consists of an optical projection-detection system with two mutually oblique axes. Light projected onto the rail is reflected and imaged onto a split photocell. A depression or rise in the reflective surface causes the axis of focus to shift and more light falls on one of the photocells. The difference in optical power between the two photocells is therefore a quantitative measure of displacement and consequently rail surface profile. The intention is to fix the device onto a bogey of a train so that measurements are taken as it moves over the track at speeds of up to 40ms-1. The relationship between vertical displacement and differential optical power is dependent on the shape and size of the spot. It is nonlinear for a circular spot and imperfections spanning less than 1 mm vertically can be resolved. Vertical displacements greater than 5mm cannot be detected solely using the difference in optical power and the nett power falling on both photocells must be considered in addition. An investigation into the reflective properties of rail lead to the optics being configured for specular reflection. The angles of incidence and reflection are optimally set at 35°. In addition, the use of an infra-red source is found to be an added advantage.
- ItemOpen AccessThe measurement of flow velocity distribution(1981) Dann, Michael Stephen; Greene, J RA method for the improvement of the range and accuracy achieved by the cross-correlation flowmeter is investigated. The principles of the flowmeter operation and fundamental digital signal processing techniques are reviewed. The process of the Fourier transform deconvolution is investigated. Computer simulation of the flow system is described and is shown to require impractical amounts of computer time to achieve the necessary averaging times. Consequently, correlation and velocity profile measurements are made from an experimental flow rig. A waveform analysis program is used to analyse these measurements. The Fourier transform deconvolution is shown in this case to have poor noise immunity. For this reason, an alternative method of Bayesian deconvolution is investigated. The correlation functions measured from the experimental flow rig are deconvolved using the Bayesian deconvolution algorithm. The resulting transit time distribution is shown to converge to the transit time distribution obtained from the velocity profile measurements. From an analysis of the flow signals the velocity distribution of the flow may thus be found.
- ItemOpen AccessSOFTWIRE : an interactive, computer-based system for enhancing learning in electrical engineering using simulation(1987) Geerdts, C D; Greene, J RThe aim of the dissertation is to explore a method of enhancing learning in the Electrical Engineering curriculum, which effectively exploits the computer. The different modes used in computer-based learning are discussed and compared, and the conclusion reached that simulation provides the best basis for a learning enhancement system. It has the ability to facilitate demonstration of basic concepts, learning of estimation, modular system design, and the use of models in engineering, and group work. It is a useful complement to laboratory work. It also enhances motivation and interest. A system is motivated loosely based on the architecture of the analogue computer, but tailored for an educational environment by being interactive, simple-to-use, adaptable and extendable by the tutor, and carrying a wide variety of educationally valuable inbuilt functions. The system proposed, SOFTWIRE, consists of a software package, a hardware laboratory interface as well as a broad approach to strategy, based on basic Learning Theory. Design issues relating to the hardware, software, and implementation of pedagogic strategy are discussed. Some examples of the use of the system are given. Thus both the broad and specific approach are covered. Some of the modes of useage discussed are demonstrations, design problems involving the construction of simulations in SOFTWIRE's language, interacting with simulations to experience them and to see the effects of parameter and structure changes, and learning about modelling and its relation to the real world. Evaluation is discussed, especially in relation to SOFTWIRE. Conclusions are drawn, and suggestions made about future development of the SOFTWIRE system, as well as relevant trends in computer education.
- ItemOpen AccessSPCATS (Sound Programme Circuit Automatic Test-Set)(1987) Galloway, A J; Greene, J R