Proactive selection of metaheuristics based on knowledge of previous results
Alejandro Rosete Suárez, Mailyn Moreno Espino · 2013
This paper presents two proactive algorithms that act as meta-metaheuristic agents: they decide which metaheuristic will be used to solve a new problem.These meta-metaheuristic agents operate in the environment of iterative work on optimizing problems, with the goal of selecting good metaheuristics to solve new problems.The information about previous results is converted into explicit knowledge that is used by the meta-metaheuristic agents to decide the most adequate metaheuristics.This proactive decision is based on a fuzzy vector that describes each problem.The proposal has been validated through experimentation with 28 functions on binary strings.