event
Thursday 13 Oct 2011: CS seminar - Doing More with Less: Improving the Performance of Genetic Programming
Colin Johnson - University of Kent
Harrison 203 15:00-16:00
Genetic Programming (GP) is a powerful technology for the automatic
creation of computer programs and other forms of executable structures
(circuits, algorithms, protocols). GP consists of the application of
biologically-inspired search techniques, such as genetic algorithms, to
the search space of computer programs. In this talk I would like to talk
about three ways in which we can improve the performance of GP:
(1) Treating each individual as a source of multiple programs.
(2) Preventing crossover between parents that produce children that
are semantically identical to the parent.
(3) Doing multiple runs of GP where each run builds on the successes
of the previous one.
Biography: Dr Colin Johnson is a Senior Lecturer in the School of Computing at the
University of Kent in Canterbury. His research interests include
bio-inspired computing and computational methods in medicine and
biology. He is also active in popularising science for the general
public, regularly giving talks about science and running a science and
technology discussion group in Canterbury.