Concentration bounds for temporal difference learning with linear function approximation: The case of batch data and uniform sampling
This work provides an efficient, low-complexity alternative for policy evaluation in big data settings, though it is incremental as it builds on existing LSTD and SGD methods.
The paper tackles the computational complexity of policy evaluation in reinforcement learning by proposing a stochastic approximation method that uniformly samples from a dataset, achieving an O(d) improvement over LSTD while maintaining convergence rates. It demonstrates practicality in traffic signal control and news recommendation applications.
We propose a stochastic approximation (SA) based method with randomization of samples for policy evaluation using the least squares temporal difference (LSTD) algorithm. Our proposed scheme is equivalent to running regular temporal difference learning with linear function approximation, albeit with samples picked uniformly from a given dataset. Our method results in an $O(d)$ improvement in complexity in comparison to LSTD, where $d$ is the dimension of the data. We provide non-asymptotic bounds for our proposed method, both in high probability and in expectation, under the assumption that the matrix underlying the LSTD solution is positive definite. The latter assumption can be easily satisfied for the pathwise LSTD variant proposed in [23]. Moreover, we also establish that using our method in place of LSTD does not impact the rate of convergence of the approximate value function to the true value function. These rate results coupled with the low computational complexity of our method make it attractive for implementation in big data settings, where $d$ is large. A similar low-complexity alternative for least squares regression is well-known as the stochastic gradient descent (SGD) algorithm. We provide finite-time bounds for SGD. We demonstrate the practicality of our method as an efficient alternative for pathwise LSTD empirically by combining it with the least squares policy iteration (LSPI) algorithm in a traffic signal control application. We also conduct another set of experiments that combines the SA based low-complexity variant for least squares regression with the LinUCB algorithm for contextual bandits, using the large scale news recommendation dataset from Yahoo.