Handling Uncertainty with a Real-coded EA

Presence of uncertainty in the search environment of Evolutionary algorithms (EA) interferes with the evaluation and the selection process of EA and adversely affects the performance of the algorithm. Presence of noise also means fitness function can not be evaluated and it has to be estimated instead. Of the various approaches which been tried to handle uncertainty in EA search environment, the more familiar approaches are: introduction of diversity (hyper mutation, random immigrants, special operators); and incorporation of memory of the past (diploidy, case based memory) [6]. In [2], we proposed a method, DPGA (distributed population evolutionary algorithm) that uses a distributed population architecture to simulate a distributed, self-adaptive memory of the solution space. Local regression is used in each sub-population to estimate the fitness. In the current research, we further investigate performance of DPGA for noisy fitness function i.e. fitness of any solution is altered by the addition of a noise term . 'Noisy' versions of few standard benchmark problems have been considered in the simulation runs of the DPGA algorithm.