Model-Based Optimization in Dynamic Environments

Model-based approaches to optimization, such as estimation-of-distribution algorithms (EDAs), ant colony optimization (ACO), and the cross-entropy method, can be very effective techniques when combined properly with local search. Building on work that considers how ACO performs in stochastic combinatorial optimization problems, this project will seek to extend model-based techniques to account for model errors that would arise under certain forms of online disturbance to decision variables, constraints or fitness evaluations.

PhD candidates with an interest in machine learning and optimization should apply.