D. Rodríguez Penas, P. González Gómez, R. Doallo Biempica, J. Rodríguez Banga
Challenging optimization problems arise during the development of large-scale kinetic models within computational Systems Biology, due to the complexity of the mathematical structures used.
Global optimization methods, such as metaheuristics, can help to solve these NP-hard problems, returning solutions near to global optimum, but with an expensive computational cost. Furthermore, there is no general heuristic to obtain good solutions for all sorts of case studies.
Recent works have studied how to improve metaheuristics with the goal to reach better results in less time, using the features of the current computational infrastructures. Thus, we propose a self-adaptive distributed multimethod: multiple optimization methods (DE, eSS, others) are performed concurrently, sharing solutions between them, and also changing their configuration to the most successful methods during the runtime. The aim is to obtain a very robust optimization tool, improving the quality of the solution reached.
Palabras clave / Keywords: cooperative metaheuristic, distributed computing, global optimization
Programado
Sesión J05 Heurísticas y Metaheurísticas
31 de mayo de 2018 10:20
Sala 4
