Guiding for Associative Learning : How to Shape Artificial Dynamic Cognition ?

Je n'ai pas de nouvelles de la date de parution de ces actes. La conférence est passée, les organisateurs ont confirmés que les papiers sortiront dans deux numéros de LNAI, en attendant, je souhaitais mettre notre travail en ligne pour qu'il soit diffusé ...International audienceThis paper describes an evolutionary robotics experiment, which aims at showing the possibility of learning by guidance in a dynamic cognition perspective. Our model relies on Continuous Time Recurrent Neural Networks and Hebbian plasticity. The agents have the ability to be guided by stimuli and we study the influence of a guidance on their external behavior and internal dynamic when faced with other stimuli. The article develops the experiment and presents some results on the dynamic of the systems