Improved Adaptive Differential Evolution Algorithm With External Archive

Depending on the complexity of the optimization problem, the performance of differential evolution (DE) algorithm is quite sensitive to the choice of mutation and crossover strategies and their associated control parameters. To obtain optimal performance, while avoiding time consuming parameter tuning, different adaptive and self-adaptive techniques that can update the strategies and/or the parameters during the evolution have been proposed. Adaptive differential evolution with optional archive (JADE) is one of the popular adaptive algorithms that perform well on most of the optimization problems. Motivated by the performance of the JADE algorithm, this paper presents an improved adaptive differential evolution algorithm with external archive (iJADE). Unlike the optional archive in JADE, iJADE algorithm employs an external archive which is updated every generation by tournament selection to incorporate the parents which cannot progress to the next generation. In addition, iJADE uses an ensemble of two crossover strategies, binomial and exponential, instead of a single crossover strategy as in JADE. The performance of the algorithm is evaluated on a set of 16 bound-constrained problems designed for Conference on Evolutionary Computation (CEC) 2005 and is compared with JADE algorithm.