Kinematic State Abstraction and Provably Efficient Rich-Observation Reinforcement Learning.

Dipendra Misra, Mikael Henaff, Akshay Krishnamurthy, John Langford

Browse the full ICML paper archive.