2005 · 245 citations · 17 references
ComputingComputer ScienceAlgorithmsArtificial IntelligenceMachine LearningGraph MiningGraph-based LearningPolicy LearningReinforcement LearningReinforcement Learning ModelsReinforcement Learning TheoryState Space ExplorationState Transition GraphsState-space Structure
We present a new subgoal-based method for automatically creating useful skills in reinforcement learning. Our method identifies subgoals by partitioning local state transition graphs---those that are constructed using only the most recent experiences of the agent. The local scope of our subgoal discovery method allows it to successfully identify the type of subgoals we seek---states that lie between two densely-connected regions of the state space while producing an algorithm with low computational cost.
17
Network Flows: Theory, Algorithms, and Applications.
David K. Smith, Ravindra K. Ahuja, Thomas L. Magnanti et al. · Journal of the Operational Research Society · 1994 · 8.2K citations · Full text