Genetic Algorithms and Programming

3 credits · Graduate

The concept of stochastic search algorithms is introduced by way of answers to the generic machine learning requirements: representation, goal state, and credit assignment. Schema theory is introduced as an underlying model for evolutionary problem solving. The significance of assuming different representations is investigated through various case studies. Different forms of 'goal state' are investigated, including multi-objective models and co-evolution are investigated in some detail and demonstrated to provide the basis for problem decomposition, game behavior design and computational efficiency.