Extending multi-objective differential evolution for optimization in presence of noise

The paper aims at designing new strategies to extend the selection step of traditional Differential Evolution for Multi-objective Optimization algorithm to proficiently obtain Pareto-optimal solutions in presence of noise. The first strategy, referred to as adaptive selection of sample size, is employed to balance the trade-off between accurate fitness estimate and computational complexity. The second strategy is concerned with determining defuzzified centroid value of the noisy fitness samples, instead of their conventional averaging, as the fitness measure of the trial solutions. The third extension is concerned with the introduction of a probabilistic Pareto ranking strategy to tarnish the detrimental effect of noise incurred in deterministic selection of traditional algorithms. The fourth strategy attempts to extend Goldberg’s approach to examine possible placement of a slightly inferior solution in the optimal Pareto front using a more statistically viable comparator. Finally, to ensure the diversity in distribution of quality solutions in the noisy fitness landscapes, a new selection criterion induced by the crowding distance measure and the probability of dominance is formulated. Experiments undertaken to study the performance of the extended algorithm reveal that the extended algorithm outperforms its competitors with respect to four performance metrics, when examined on a test-suite of 23 standard benchmarks with additive noise of three statistical distributions.