The evolutionary computation group studies the process of evolving better solutions from a population of potentially good solutions. This is a process similar to a farmer breeding a better cow from a herd of cows. These techniques are useful for complex problems for which no known efficient (polynomial) algorithms exist or where the search space is immense.
Specifically this group has looked at several techniques of breeding protein classifiers for biological problems and training teams of robots to cooperate. The group also works on theory to understand how subtle interactions in a problem representation called epistasis that makes the problems more difficult for these general optimizers.
For more information visit the Initiative for Bioinformatics and Evolutionary Studies (IBEST) web site.
Participating Faculty: Robert Heckendorn and Terry Soule