Les oiseaux picorant artificiels : une nouvelle méta-heuristique inspirée du comportement des pigeons

Many optimization algorithms and metaheuristics have been inspired by nature, e.g. by the social behavior of honey bees for Artificial Bee Colony (ABC). These algorithms often permit solving a wide range of optimization problems. In this paper, we present Artificial Feeding Birds (AFB), a new metaheuristic inspired by the behavior of pigeons searching for food. AFB is very simple, yet efficient, and easy to adapt to various optimization problems. We apply it to unconstrained global nonlinear optimization (including the training of artificial neural networks) and to the resolution of traveling salesman problem. We also show that AFB gives results equivalent or better than ABC on several benchmark functions. Finally, we discuss the differences between the two algorithms and the choice of inspiration sources from nature.