Genetic algorithm and data visualization

• Main Author: Haroun, Paul.
• Other Authors: Devroye, Luc (advisor)
• Format: Others
• Language: en
• Published: McGill University 1997
• Subjects: Computer Science.
• Online Access: http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=27526

The genetic algorithm (GA) is an adaptive search algorithm used in high-dimensional optimization problems where analytical methods are either unavailable or impractical. The GA is stochastic by nature. It requires a well-tuned set of control parameters in order to perform well. The values of the control parameters depend on the nature of the problem. Finding the appropriate set of values for these parameters can be quite difficult. Using computer graphics, alongside theoretical analysis, to visualize the data generated by the GA aids in the study of the behavior of the genetic algorithm and helps in the fine-tuning of the parameters. In this thesis, the GA is described using John Holland's formal framework. An extension to this framework to encompass visual representations using geometric and organizational modeling is defined and an implementation, called PHGA, of both the genetic algorithm and its visual representation is discussed.