Policy Gradients Incorporating the Future
This addresses the problem of future reasoning in RL for agents operating in stochastic or partially observable settings, representing an incremental improvement by integrating future information into existing policy gradient methods.
The paper tackles the challenge of reasoning about future outcomes in reinforcement learning by introducing a method that allows agents to observe actual future trajectories during training while using an information bottleneck to prevent over-reliance on this privileged information. The result is that PGIF achieves higher reward faster in various RL domains, including sparse-reward and partially observable environments.
Reasoning about the future -- understanding how decisions in the present time affect outcomes in the future -- is one of the central challenges for reinforcement learning (RL), especially in highly-stochastic or partially observable environments. While predicting the future directly is hard, in this work we introduce a method that allows an agent to "look into the future" without explicitly predicting it. Namely, we propose to allow an agent, during its training on past experience, to observe what \emph{actually} happened in the future at that time, while enforcing an information bottleneck to avoid the agent overly relying on this privileged information. This gives our agent the opportunity to utilize rich and useful information about the future trajectory dynamics in addition to the present. Our method, Policy Gradients Incorporating the Future (PGIF), is easy to implement and versatile, being applicable to virtually any policy gradient algorithm. We apply our proposed method to a number of off-the-shelf RL algorithms and show that PGIF is able to achieve higher reward faster in a variety of online and offline RL domains, as well as sparse-reward and partially observable environments.