Comparaison de l'optimisation par colonies de fourmis et d'une recherche taboue réactive sur des problèmes d'appariement de graphes

Many applications involve measuring object similarity. When the objects are modeled by graphs, measuring the similarity turn into a graph matching problem, i.e., into the search for a best matching between the graphs vertices. In this paper, we address the problem of computing a graph similarity measure based on a multivalent graph matching. This measure has been shown generic in the sense that other well-known graph similarity measures can be viewed as special cases of it. We propose and compare two di erent kinds of algorithms : an Ant Colony Optimization based algorithm and a Reactive Tabu Local Search. We compare the e ciency of these two algorithms on two di erent kinds of di cult graph matching problems and we show that they obtain complementary results.