Thursday 17 Oct 2013: Metaheuristics for Optimisation in Noisy Environments
Prof Juergen Branke - Warwick Business School
Harrison 101 15:00-16:00
Most publications on optimisation assume that the objective function is clearly defined and deterministic. In many practical applications, however, evaluations of solutions are subject to noise, for example if solutions are evaluated by running a stochastic simulation, or when evaluation involves conducting physical experiments. This obviously impacts an algorithm's ability to compare the quality of solutions and identify the better ones, which is a key component of metaheuristcs such as evolutionary algorithms and simulated annealing.
This talk discusses various ways to adapt metaheuristics to address the particular challenges of optimisation in noisy environments.