Most learning algorithms are not invariant to the scale of the signal that is being approximated. We propose to adaptively normalize the targets used in the learning updates. This is important in value-based reinforcement learning, where the magnitude of appropriate value approximations can change over time when we update the policy of behavior. Our main motivation is prior work on learning to play Atari games, where the rewards were clipped to a predetermined range. This clipping facilitates learning across many different games with a single learning algorithm, but a clipped reward function can result in qualitatively different behavior. Using adaptive normalization we can remove this domain-specific heuristic without diminishing overall performance.

Given the impressive results of deepreinforcement learning, weargueforaneedtomoreclearly understand the challenges inthis setting.

One of the most effectivecontinuous deep reinforcement learning algorithms is normalized advantage functions (NAF).

One of the phenomena of interest is that Q-learning (Watkins, 1989) is known to suffer from overestimation issues, since it takes a maximum operator overaset ofestimated action-values.

Neural Information Processing Systems http://nips.cc/

In particular, we look at commonalities and differences between parametric models and experience replay.

Credit assignment, i.e. determining how to correctly associate delayed rewards with states or state-action pairs, is a crucial problem for reinforcement learning (RL) agents ( Sutton and Barto, 2018).