Regresar

Impacto de la auto-adaptación en ambientes dinámicos con frecuencia de cambio variable

Abstract:

Several decision scenarios can be modeled as dynamic optimization problems (DOPs), which have been tackled by metaheuristics techniques over the last years. However, so far, most related works assume that changes occur every equal time periods, which may be rather idealistic in real-world scenarios. In contrast, DOPs with variable change frequency (DOPVCFs) impose as main challenge to the algorithm: how to adapt to different environments during the run, as fast as possible. In that sense, self-adaptation is parameter control technique with remarkable success in complex scenarios. So, in the present work we aim to analyze the impact of self-adaptation in solving DOPVCFs. To achieve this, we have designed an experimental study by considering a recently proposed self-adaptive technique over several test scenarios. The results confirm that self-adaptation has not only a positive, but also significant impact in the algorithm performance.