Reinforcement learning for autonomic network repair

We report on our efforts to formulate autonomic network repair as a reinforcement-learning problem. Our implemented system is able to learn to efficiently restore network connectivity after a failure.