Backplay: "Man muss immer umkehren"
This addresses sample efficiency for RL practitioners in sparse-reward environments, though it is incremental as it builds on existing curriculum learning approaches.
The paper tackles the problem of low sample efficiency in model-free reinforcement learning with sparse rewards by introducing Backplay, a method that uses a single demonstration to create a curriculum starting near the end and moving backward, resulting in improved training speed in grid worlds and a complex game like Pommerman compared to other methods.
Model-free reinforcement learning (RL) requires a large number of trials to learn a good policy, especially in environments with sparse rewards. We explore a method to improve the sample efficiency when we have access to demonstrations. Our approach, Backplay, uses a single demonstration to construct a curriculum for a given task. Rather than starting each training episode in the environment's fixed initial state, we start the agent near the end of the demonstration and move the starting point backwards during the course of training until we reach the initial state. Our contributions are that we analytically characterize the types of environments where Backplay can improve training speed, demonstrate the effectiveness of Backplay both in large grid worlds and a complex four player zero-sum game (Pommerman), and show that Backplay compares favorably to other competitive methods known to improve sample efficiency. This includes reward shaping, behavioral cloning, and reverse curriculum generation.