 |
 |
FULL-TIME RESEARCH STUDENTSHIP IN GENETIC ALGORITHMS Applications are invited for a full-time research studentship at the Open University, tenable from 1 October 2000. Genetic Algorithms for Robust Optimisation Primary supervisor: Dr AJ Hirst, [log in to unmask] Secondary supervisor: Dr AA Hopgood, [log in to unmask] DESCRIPTION: Genetic Algorithms (GAs) use the principles of Natural and Artificial Selection to find solutions to computational problems. The aim of the project is to determine the extent to which this novel computational method can be applied to parametric design problems to automatically identify solutions with desired sensitivity characteristics. This is of importance not only for the theoretical understanding of genetic algorithms, but also adds to the 'use- value' of the process for users of the technology in industrial design applications. The project will consider the use of a hybrid process that combines a genetic algorithm with a local search procedure in a novel way. Typically, local search is to find the best local neighbour of an individual in some search space, a process akin to 'learning' in the sense of an individual improving its performance on some task. However, by using local search to identify the worst neighbour (a process akin to 'fault induction') the evolutionary process may be able to identify 'robust' solutions to a specified problem.. The broad aim of the project is to identify the sense in which evolution 'optimises' individuals and populations. Previous work seems to suggest that evolution is a 'robust optimiser' rather than a global optimiser, in the sense that solutions are found that are tolerant to perturbation. After formally clarifying the notion of evolutionary optimisation, several questions will be addressed: * if evolution is a robust optimiser, can operator sets be chosen such that solutions with a predictable sensitivity can be automatically located by the evolutionary process? * how do learning and fault tolerance affect the sensitivity of discovered solutions? * how well does evolution compare with other optimisation methods? * can population dynamics be used to demonstrate a path between a current configuration and a desired configuration using operators based on operations that may be applied to a configuration in the real world? * to what extent are population dynamics and the dynamics of a changing selective environment correlated? In the first instance, the target domain will be the evolution of parametric designs, in particular analogue electronic circuits. Although there are many evolutionary algorithm code libraries publicly available, very few of them have a well developed graphical front end. An initial requirement of the project will be the adoption of one of the currently offered code libraries and the development of a graphical interface for its use. This will make work easier in the long run. A secondary objective is that the interface is useful not only in pursuance of the experiments required by the project, but would also afford a front end appropriate for use by students in an OU course with a GA project component. The successful applicant will undergo a complete research-training programme tailored specifically to their needs within a flourishing intellectual and research environment. The studentship will be based in the IntelligentComputer Systems (ICS) Research Group (http://www-tec.open.ac.uk/tel/res/isrg/ )within Telematics Department (the http:///www-tec.open.ac.uk/tel). OU FULL-TIME STUDENTSHIPS: The Open University's full-time students are normally expected to live in the Milton Keynes area and use the University's research facilities. Formal information regarding OU studentships is available from the Research Degrees Center webpage (http://www.open.ac.uk/research-degrees/). Further information about life as a full-time OU postgrad. is available from the Postgraduate Students' Society webpage (http://www.open.ac.uk/pgss/). HOW TO APPLY Informal enquiries may be directed to Tony Hirst at [log in to unmask] or tel. 01908 652789. Applicants should have or expect to gain in 2000 a good honours degree or Masters degree in a relevant subject. Experience of programming in C/C++ will be an advantage. The studentship will be based in the Intelligent Computer Systems (ICS) Research Group (http://www-tec.open.ac.uk/tel/res/isrg/) within the Telematics Department (http:///www-tec.open.ac.uk/tel). For a prospectus please phone +44 (0) 1908 653924 or 653355 or email [log in to unmask] An online application form is available in MS-word format from http://www.open.ac.uk/research-degrees/rdegform.doc Alternatively, write to the address below or visit the Faculty of Technology recruitment website at: http://www-tec.open.ac.uk/deanery/recruit.html The closing date for applications is 31st May, 2000. Equal Opportunity is University Policy. Department of Telematics, The Open University, Walton Hall, Milton Keynes, MK7 6AA. -- ---------------------------------------------------- Tony Hirst mailto:[log in to unmask] Dept. of Telematics, Faculty of Technology Open University, Walton Hall, Milton Keynes, MK7 6AA, UK Tel: +44 (0)19086 52789 http://www-tec.open.ac.uk/tel/people/hirst/tony/ ---------------------------------------------------- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%