Título: Recent Hybrid Metaheuristics for Combinatorial Optimization

Ponente: Christian Blum (Artificial Intelligence Research Institute (IIIA-CSIC))

Organizador: Javier Alcaraz

Date: Lunes 21 de noviembre de 2022 a las 12:30 horas

Lugar: Thinking Lab

Abstract: In this talk, I will explain the main ideas behind two of our recent algorithmic developments from the field of hybrid metaheuristics, that is, combinations of metaheuristics with other techniques for optimization. Both algorithms can generally be applied to any combinatorial optimization problem. For different purposes, they internally solve sub-instances of the tackled problem instances by using, for example, MILP solvers or more specialized exact techniques that have been developed for the considered optimization problems. The first example concerns an extension of the metaheuristic ant colony optimization by a negative learning mechanism. The second example is about an award-winning algorithm known as construct, merge, solve & adapt (CMSA). Both algorithms will be introduced by means of examples. Finally, we will present a new web application of our graphical tool called «search trajectory networks (STNs)» for visualizing and analyzing algorithm behaviour.

Categories: Novedades