Regularization and Feature Selection in Least Squares Temporal-Difference Learning

We consider the task of reinforcement learning with linear value function approximation. Temporal difference algorithms, and in particular the Least-Squares Temporal Difference (LSTD) algorithm, provi