arXiv (Cornell University) · 2020 · 19 citations · 33 references
Open access
ComputingComputer ScienceAlgorithmsArtificial IntelligenceMachine LearningOptimal ControlPolicy LearningReinforcement LearningReinforcement Learning FoundationsValue IterationApproximate Value FunctionsApproximate Value IterationAverage Reward Q-learningError PropagationQ-learningRegularization
Recent Reinforcement Learning (RL) algorithms making use of Kullback-Leibler (KL) regularization as a core component have shown outstanding performance. Yet, only little is understood theoretically about why KL regularization helps, so far. We study KL regularization within an approximate value iteration scheme and show that it implicitly averages q-values. Leveraging this insight, we provide a very strong performance bound, the very first to combine two desirable aspects: a linear dependency to the horizon (instead of quadratic) and an error propagation term involving an averaging effect of the estimation errors (instead of an accumulation effect). We also study the more general case of an additional entropy regularizer. The resulting abstract scheme encompasses many existing RL algorithms. Some of our assumptions do not hold with neural networks, so we complement this theoretical analysis with an extensive empirical study.
33
Human-level control through deep reinforcement learning
Volodymyr Mnih, Koray Kavukcuoglu, David Silver et al. · Nature · 2015 · 30K citations
Trust Region Policy Optimization
John Schulman, Sergey Levine, Philipp Moritz et al. · arXiv (Cornell University) · 2015 · 3.1K citations · Full text