Performance of network crossover on NK landscapes and spin glasses

This paper describes a network crossover operator based on knowledge gathered from either prior problem-specific knowledge or linkage learning methods such as estimation of distribution algorithms (EDAs). This operator can be used in a genetic algorithm (GA) to incorporate linkage in recombination. The performance of GA with network crossover is compared to that of GA with uniform crossover and the hierarchical Bayesian optimization algorithm (hBOA) on 2D Ising spin glasses, NK landscapes, and SK spin glasses. The results are analyzed and discussed.