Bootstrapped Reward Shaping
This work addresses the problem of slow training in sparse-reward domains for reinforcement learning practitioners, offering an incremental improvement over existing reward shaping methods.
The paper tackles the challenge of sparse rewards in reinforcement learning by introducing Bootstrapped Reward Shaping (BSRS), which uses the agent's state-value estimate as a potential function to provide denser rewards without altering the optimal policy, resulting in improved training speed on the Atari suite.
In reinforcement learning, especially in sparse-reward domains, many environment steps are required to observe reward information. In order to increase the frequency of such observations, "potential-based reward shaping" (PBRS) has been proposed as a method of providing a more dense reward signal while leaving the optimal policy invariant. However, the required "potential function" must be carefully designed with task-dependent knowledge to not deter training performance. In this work, we propose a "bootstrapped" method of reward shaping, termed BSRS, in which the agent's current estimate of the state-value function acts as the potential function for PBRS. We provide convergence proofs for the tabular setting, give insights into training dynamics for deep RL, and show that the proposed method improves training speed in the Atari suite.