Data-Efficient Policy Evaluation Through Behavior Policy Search
This work addresses data efficiency in policy evaluation for reinforcement learning, offering a method to improve accuracy with less data, though it is incremental as it builds on existing unbiased estimation techniques.
The paper tackles the problem of evaluating a policy in Markov decision processes by showing that using data from a different behavior policy can produce unbiased estimates with lower mean squared error than standard deployment, and it proposes a behavior policy search algorithm that empirically reduces this error.
We consider the task of evaluating a policy for a Markov decision process (MDP). The standard unbiased technique for evaluating a policy is to deploy the policy and observe its performance. We show that the data collected from deploying a different policy, commonly called the behavior policy, can be used to produce unbiased estimates with lower mean squared error than this standard technique. We derive an analytic expression for the optimal behavior policy --- the behavior policy that minimizes the mean squared error of the resulting estimates. Because this expression depends on terms that are unknown in practice, we propose a novel policy evaluation sub-problem, behavior policy search: searching for a behavior policy that reduces mean squared error. We present a behavior policy search algorithm and empirically demonstrate its effectiveness in lowering the mean squared error of policy performance estimates.