Adaptive routing process by deflection with training by reinforcement

摘要

The invention concerns a management process for a network of routers based on the technique of training by reinforcement in which priority is given to objects already present in the network over those which wish to enter.