I see that discounted reward reinforcement learning has been extensively studied in the literature. However, the average reward metric receives less attention, and it looks like algorithms for this metric (R-learning, H-learning, SMART, etc.) are less than the discount metric. Could you suggest any algorithms for average reward reinforcement learning in continuous or general state-action space?

  • $\begingroup$ can you please add the references to the papers? $\endgroup$
    – Alberto
    Oct 11, 2023 at 12:29


You must log in to answer this question.