When Hillclimbers Beat Genetic Algorithms in Multimodal Optimization
Fernando Graça Lobo, Mosab Bazargani · 2015
We show that multistart next ascent hillclimbing compares favourably to crowding-based genetic algorithms when solving instances of the multimodal problem generator. We conjecture that it is unlikely that any practical evolutionary algorithm is capable of solving this type of problem instances faster than the multistart hillclimbing strategy.