Schwerpunkt: Learning and Self-organization of Behavior
Information in the Perception and Action of Embodied Agents
Zusammenfassung:
The use of Shannon’s information theory in Artificial Intelligence scenarios is increasingly popular. We outline a project framework in which we use information theory to analyse and structure the dynamics of the perception and action of embodied agents. It encompasses abstract proof-of-principle scenarios as well as implementations operating on robotic hardware. The results obtained in this project framework indicate that methods based on information theory provide a promising approach towards the development of Artificial Intelligence and towards bridging the gap to biologically plausible mechanisms for intelligence.
Autoren:
Daniel Polani, Chrystopher L. Nehaniv, Alexander S. Klyubin, Lars Olsson
