Proximal Gradient Temporal Difference Learning: Stable Reinforcement Learning with Polynomial Sample Complexity
This work addresses the challenge of off-policy learning in reinforcement learning with improved stability and efficiency, though it is incremental as it builds on existing GTD methods.
The paper tackles the problem of stable reinforcement learning with finite-sample guarantees by introducing proximal gradient temporal difference learning, showing that the GTD family of algorithms achieves polynomial sample complexity and may be preferred over least squares TD methods due to linear complexity.
In this paper, we introduce proximal gradient temporal difference learning, which provides a principled way of designing and analyzing true stochastic gradient temporal difference learning algorithms. We show how gradient TD (GTD) reinforcement learning methods can be formally derived, not by starting from their original objective functions, as previously attempted, but rather from a primal-dual saddle-point objective function. We also conduct a saddle-point error analysis to obtain finite-sample bounds on their performance. Previous analyses of this class of algorithms use stochastic approximation techniques to prove asymptotic convergence, and do not provide any finite-sample analysis. We also propose an accelerated algorithm, called GTD2-MP, that uses proximal ``mirror maps'' to yield an improved convergence rate. The results of our theoretical analysis imply that the GTD family of algorithms are comparable and may indeed be preferred over existing least squares TD methods for off-policy learning, due to their linear complexity. We provide experimental results showing the improved performance of our accelerated gradient TD methods.