BlendRL: A Framework for Merging Symbolic and Neural Policy Learning
This addresses the limitation in RL agents' capabilities for flexible low-level reactions and interpretable reasoning, though it appears incremental as it builds on existing neuro-symbolic approaches.
The paper tackles the problem of disjointed symbolic and neural policy learning in reinforcement learning by introducing BlendRL, a neuro-symbolic RL framework that integrates both paradigms, resulting in agents outperforming neural and symbolic baselines in Atari environments and showing robustness to environmental changes.
Humans can leverage both symbolic reasoning and intuitive reactions. In contrast, reinforcement learning policies are typically encoded in either opaque systems like neural networks or symbolic systems that rely on predefined symbols and rules. This disjointed approach severely limits the agents' capabilities, as they often lack either the flexible low-level reaction characteristic of neural agents or the interpretable reasoning of symbolic agents. To overcome this challenge, we introduce BlendRL, a neuro-symbolic RL framework that harmoniously integrates both paradigms within RL agents that use mixtures of both logic and neural policies. We empirically demonstrate that BlendRL agents outperform both neural and symbolic baselines in standard Atari environments, and showcase their robustness to environmental changes. Additionally, we analyze the interaction between neural and symbolic policies, illustrating how their hybrid use helps agents overcome each other's limitations.