Cara MacNish
University of Western Australia
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Cara MacNish.
IEEE Transactions on Evolutionary Computation | 2004
Andrew Czarn; Cara MacNish; Kaipillil Vijayan; Berwin A. Turlach; Ritu Gupta
Genetic algorithms have been extensively used and studied in computer science, yet there is no generally accepted methodology for exploring which parameters significantly affect performance, whether there is any interaction between parameters, and how performance varies with respect to changes in parameters. This paper presents a rigorous yet practical statistical methodology for the exploratory study of genetic and other adaptive algorithms. This methodology addresses the issues of experimental design, blocking, power calculations, and response curve analysis. It details how statistical analysis may assist the investigator along the exploratory pathway. As a demonstration of our methodology, we describe case studies using four well-known test functions. We find that the effect upon performance of crossover is pre-dominantly linear, while the effect of mutation is predominantly quadratic. Higher order effects are noted but contribute less to overall behavior. In the case of crossover, both positive and negative gradients are found suggesting the use of a maximum crossover rate for some problems and its exclusion for others. For mutation, optimal rates appear higher compared with earlier recommendations in the literature, while supporting more recent work. The significance of interaction and the best values for crossover and mutation are problem specific.
international conference on requirements engineering | 1996
Miles Osborne; Cara MacNish
Ambiguity in requirement specifications causes numerous problems; for example in defining customer/supplier contracts, ensuring the integrity of safety-critical systems, and analysing the implications of system change requests. A direct appeal to formal specification has not solved these problems, partly because of the restrictiveness and back of habitability of formal languages. An alternative approach, described in the paper, is to use natural language processing (NLP) techniques to aid the development of formal descriptions from requirements expressed in controlled natural language. While many problems in NLP remain unsolved, we show that suitable extensions to existing tools provide a useful platform for detecting and resolving ambiguities. Our system is demonstrated through a case-study on a simple requirements specification.
Connection Science | 2007
Cara MacNish
Randomised population-based algorithms, such as evolutionary, genetic and swarm-based algorithms, and their hybrids with traditional search techniques, have proven successful and robust on many difficult real-valued optimisation problems. This success, along with the readily applicable nature of these techniques, has led to an explosion in the number of algorithms and variants proposed. In order for the field to advance it is necessary to carry out effective comparative evaluations of these algorithms, and thereby better identify and understand those properties that lead to better performance. This paper discusses the difficulties of providing benchmarking of evolutionary and allied algorithms that is both meaningful and logistically viable. To be meaningful the benchmarking test must give a fair comparison that is free, as far as possible, from biases that favour one style of algorithm over another. To be logistically viable it must overcome the need for pairwise comparison between all the proposed algorithms. To address the first problem, we begin by attempting to identify the biases that are inherent in commonly used benchmarking functions. We then describe a suite of test problems, generated recursively as self-similar or fractal landscapes, designed to overcome these biases. For the second, we describe a server that uses web services to allow researchers to ‘plug in’ their algorithms, running on their local machines, to a central benchmarking repository.
australasian conference on computer science education | 1998
David Clark; Cara MacNish; Gordon F. Royle
This paper describeqeriences in teaching Java to Computer Science students in two Australian universities. The paper highlights some of the problems encountered in teaching Java, and some of the areas that needed careful treatment. Based on these experiences -we suggest an unorthodox ‘objects-only” approach to introducing Java to new students.
australasian joint conference on artificial intelligence | 2004
Andrew Czarn; Cara MacNish; Kaipillil Vijayan; Berwin A. Turlach
An important issue in genetic algorithms is the relationship between the difficulty of a problem and the choice of encoding Two questions remain unanswered: is their a statistically demonstrable relationship between the difficulty of a problem and the choice of encoding, and, if so, what it the actual mechanism by which this occurs? In this paper we use components of a rigorous statistical methodology to demonstrate that the choice of encoding has a real effect upon the difficulty of a problem Computer animation is then used to illustrate the actual mechanism by which this occurs.
world congress on computational intelligence | 2008
Cara MacNish; Xin Yao
Directional biases are evident in many benchmarking problems for real-valued global optimisation, as well as many of the evolutionary and allied algorithms that have been proposed for solving them. It has been shown that directional biases make some kinds of problems easier to solve for similarly biased algorithms, which can give a misleading view of algorithm performance. In this paper we study the effects of directional bias for high- dimensional optimisation problems. We show that the impact of directional bias is magnified as dimension increases, and can in some cases lead to differences in performance of many orders of magnitude. We present a new version of the classical evolutionary programming algorithm, which we call unbiased evolutionary programming (UEP), and show that it has markedly improved performance for high-dimensional optimisation.
Philosophical Magazine | 2015
Elena Pasternak; Arcady Dyskin; Maxim Esin; Ghulam Mubashar Hassan; Cara MacNish
Shear band formation and evolution is a predominant mechanism of deformation patterning in granular materials. Independent rotations of separate particles can affect the pattern formation by adding the effect of rotational degrees of freedom to the mechanism of instability. We conducted 2D physical modelling where the particles are represented by smooth steel discs. We use the digital image correlation in order to recover both displacement and independent rotation fields in the model. We performed model calibration and determine the values of mechanical parameters needed for a DEM numerical modelling. Both mono- and polydisperse particle assemblies are used. During the loading, the deformation pattern undergoes stages of shear band formation followed by its dissolution due to recompaction and particle rearrangement with the subsequent formation of multiple shear bands merging into a single one and the final dissolution. We show that while the average (over the assembly) values of the angles of disc rotations are insignificantly different from zero, the particle rotations exhibit clustering at the mesoscale (sizes larger than the particles but smaller than the whole assembly): monodisperse assemblies produce vertical columns of particles rotating the same direction; polydisperse assemblies 2D form clusters of particles with alternating rotations. Thus, particle rotations produce a structure on their own, a structure different form the ones formed by particle displacements and force chains. This can give a rise to moment chains. These emerging mesoscopic structures – not observable at the macroscale – indicate hidden aspects of ‘Cosserat behaviour’ of the particles.
international conference on image processing | 2015
Nghia V. Dinh; Ghulam Mubashar Hassan; Arcady Dyskin; Cara MacNish
Digital image correlation (DIC) is a well-known contact-less technique offering highly accurate full-field deformation measurement using grayscale images. The practical implementation of DIC is still facing many challenges, especially limitations of accuracy in measuring small displacement gradients for solids in geosciences and biomedi-cal engineering. In this paper, we introduce a novel approach in which color images are employed to enhance the performance of DIC. A complete framework for Color DIC has been proposed and tested. The results show that Color DIC performs significantly better than grayscale DIC for measurement of small strains by a factor of 2.
simulated evolution and learning | 2006
Cara MacNish
The success (and potential success) of evolutionary algorithms and their hybrids on difficult real-valued optimization problems has led to an explosion in the number of algorithms and variants proposed. This has made it difficult to definitively compare the range of algorithms proposed, and therefore to advance the field. In this paper we discuss the difficulties of providing widely available benchmarking, and present a solution that addresses these difficulties. Our solution uses automatically generated fractal landscapes, and allows users algorithms written in any language and run on any platform to be “plugged into” the benchmarking software via the web.
pacific rim international conference on artificial intelligence | 1996
Cara MacNish; Mary-Anne Williams
The ability to correctly analyse the impact of changes to system designs is an important goal in software engineering. A framework for addressing this problem has been proposed in which logical descriptions are developed alongside traditional representations. While changes to the resulting design have been considered, no formal framework for design change has been offered.