Approximately Optimal Approximate Reinforcement Learning.

Sham M. Kakade, John Langford

Browse the full ICML paper archive.