Which Experiences Are Influential for Your Agent? Policy Iteration with Turn-over Dropout
This work addresses a computational bottleneck for RL practitioners by providing a more efficient way to analyze experience influence, though it appears incremental as it builds on existing policy iteration and dropout techniques.
The paper tackles the problem of efficiently estimating the influence of individual experiences in reinforcement learning with experience replay, presenting PI+ToD as a method that uses turn-over dropout to achieve this, with experiments in MuJoCo environments demonstrating its efficiency.
In reinforcement learning (RL) with experience replay, experiences stored in a replay buffer influence the RL agent's performance. Information about the influence is valuable for various purposes, including experience cleansing and analysis. One method for estimating the influence of individual experiences is agent comparison, but it is prohibitively expensive when there is a large number of experiences. In this paper, we present PI+ToD as a method for efficiently estimating the influence of experiences. PI+ToD is a policy iteration that efficiently estimates the influence of experiences by utilizing turn-over dropout. We demonstrate the efficiency of PI+ToD with experiments in MuJoCo environments.