Optimising Evolutionary Strategies for Problems with Non-uniform Noise
For many "real world" applications of evolutionary computation, the fitness function is obscured by random noise, which interferes with the evolutionary search. Furthermore, the amount of noise (noise strength) may vary throughout the search space, further complicating matters. Previous work has generally focussed on the specific case where noise strength is constant; however, we study problems with varying noise strength. We give new algorithms specifically designed to handle such problems, show how they perform, and provide a means to automatically apply them.
This thesis makes major contributions which present a significant addition to the body of knowledge for noisy fitness functions.
This thesis is the first work specifically to examine the implications of noise strength that varies throughout the search domain for a variety of noise landscapes, and thus starts to fill a large void in the literature of noisy fitness functions.