Learning from Active Human Involvement through Proxy Value Propagation
This work addresses the challenge of AI alignment and safety by allowing humans to actively intervene in training, though it is incremental as it builds on existing reinforcement learning frameworks.
The paper tackles the problem of enabling AI agents to learn from active human interventions during training by proposing Proxy Value Propagation, a reward-free method that uses a proxy value function to propagate human intent from demonstrations to unlabeled data, resulting in agents that can solve continuous and discrete control tasks, including driving in Grand Theft Auto V, with minimal algorithm modifications.
Learning from active human involvement enables the human subject to actively intervene and demonstrate to the AI agent during training. The interaction and corrective feedback from human brings safety and AI alignment to the learning process. In this work, we propose a new reward-free active human involvement method called Proxy Value Propagation for policy optimization. Our key insight is that a proxy value function can be designed to express human intents, wherein state-action pairs in the human demonstration are labeled with high values, while those agents' actions that are intervened receive low values. Through the TD-learning framework, labeled values of demonstrated state-action pairs are further propagated to other unlabeled data generated from agents' exploration. The proxy value function thus induces a policy that faithfully emulates human behaviors. Human-in-the-loop experiments show the generality and efficiency of our method. With minimal modification to existing reinforcement learning algorithms, our method can learn to solve continuous and discrete control tasks with various human control devices, including the challenging task of driving in Grand Theft Auto V. Demo video and code are available at: https://metadriverse.github.io/pvp