Policy Learning Using Weak Supervision
This work addresses the challenge of expensive supervision in reinforcement learning and behavioral cloning for practitioners, offering a more efficient approach, though it is incremental as it builds on existing weak supervision ideas.
The paper tackles the problem of policy learning when high-quality supervision is unavailable or expensive by proposing a unified framework that uses cheap weak supervision, treating it as imperfect information from a peer agent and evaluating policies based on correlated agreement to prevent overfitting. The method shows substantial performance improvements in tasks like RL with noisy rewards and BC with weak demonstrations, particularly in complex or noisy environments.
Most existing policy learning solutions require the learning agents to receive high-quality supervision signals such as well-designed rewards in reinforcement learning (RL) or high-quality expert demonstrations in behavioral cloning (BC). These quality supervisions are usually infeasible or prohibitively expensive to obtain in practice. We aim for a unified framework that leverages the available cheap weak supervisions to perform policy learning efficiently. To handle this problem, we treat the "weak supervision" as imperfect information coming from a peer agent, and evaluate the learning agent's policy based on a "correlated agreement" with the peer agent's policy (instead of simple agreements). Our approach explicitly punishes a policy for overfitting to the weak supervision. In addition to theoretical guarantees, extensive evaluations on tasks including RL with noisy rewards, BC with weak demonstrations, and standard policy co-training show that our method leads to substantial performance improvements, especially when the complexity or the noise of the learning environments is high.