| On 2018-02-05 at 11:00:00 (Brussels Time) |
Abstract
In most cases, problem solutions can't be assigned a single cost or fitness. Due to intrinsic problems (measurement imprecision), the evaluation process (using some stochastic process as part of it) or any other factors, the fitness is a statistical variable whose model is impossible to know in advance, and might not even exist. In this talk we will talk about different ways of dealing with it in metaheuristics and specially evolutionary computation, the problems they have and new ways to deal with it properly.
Keywords
Evolutionary computation, Uncertainty
