A Necessary Condition of Convergence for Reinforcement Learning with Function Approximation.

Artur Merke, Ralf Schoknecht

Browse the full ICML paper archive.