May the same numerical optimizer be used when searching either for the best or for the worst solution to a real-world problem?

摘要: Over the last two decades numerous metaheuristics have been proposed and it seems today that nobody is able to understand, evaluate, or compare them all. In principle, optimization methods, including recently popular Evolutionary Computation Swarm Intelligence-based ones, should be developed in order solve real-world problems. Yet vast majority of are tested source papers on artificial benchmarks only, so their usefulness for various practical applications remains unverified. As a result, choosing proper method particular problem difficult task. This paper shows such choice even more complicated if one wishes, with good reason, use twice, once find best then worst solutions specific numerical problem. It often occurs either case different optimizers recommended. The above finding based testing 30 problems from CEC2011. First we 22 minimization as defined Then reverse objective function each search its maximizing solution. We also observe algorithms highly ranked average may not perform any given Rather, highest ranking achieved by methods never among poorest ones. other words, occasional winners get less attention than rare losers.