Robert Collier defended his PhD dissertation

Yesterday I attended the doctoral dissertation defense of Robert Collier at the School of Computer Science at the University of Guelph (Ontario, Canada). Robert’s dissertation is entitled “Empirically Evaluated Improvements to Genotypic Spatial Distance Measurement Approaches for the Genetic Algorithm”. The dissertation focuses on the design of a distance metric for candidate solutions in genetic algorithms, which would estimate how easy it is to get from one candidate solution to another one using mutation and crossover operators. This is a topic of central importance for visualization and theoretical study of adaptive landscapes. Until now, the Hamming distance metric was the usual (and the only available) approach to measuring distances between candidate solutions in genetic algorithms. However, since the Hamming distance metric assumes search operators that change only one or a few variables at a time, the use of this distance metric in genetic algorithms has severe limitations. Robert’s thesis tackles this obstacle with focus on visualization, although the distance metrics are crucial also for other areas, such as landscape theory. Robert’s PhD advisor was Mark Wineberg. Congratulations to the successful defense of your thesis, Robert!

Visiting the University of Guelph gave me an opportunity to give a talk on estimation of distribution algorithms with special focus on BOA and hBOA algorithms.

Comments are closed.