Mitigating Information Loss in Tree-Based Reinforcement Learning via Direct Optimization
This addresses the problem of interpretability in RL for domains requiring transparent decision-making, representing an incremental advancement by enabling gradient-based learning of decision trees within existing on-policy frameworks.
The paper tackles the challenge of learning interpretable symbolic policies in on-policy reinforcement learning by introducing SYMPOL, a method that integrates tree-based models with policy gradient methods, achieving superior performance and interpretability over alternative tree-based RL approaches on benchmark tasks.
Reinforcement learning (RL) has seen significant success across various domains, but its adoption is often limited by the black-box nature of neural network policies, making them difficult to interpret. In contrast, symbolic policies allow representing decision-making strategies in a compact and interpretable way. However, learning symbolic policies directly within on-policy methods remains challenging. In this paper, we introduce SYMPOL, a novel method for SYMbolic tree-based on-POLicy RL. SYMPOL employs a tree-based model integrated with a policy gradient method, enabling the agent to learn and adapt its actions while maintaining a high level of interpretability. We evaluate SYMPOL on a set of benchmark RL tasks, demonstrating its superiority over alternative tree-based RL approaches in terms of performance and interpretability. Unlike existing methods, it enables gradient-based, end-to-end learning of interpretable, axis-aligned decision trees within standard on-policy RL algorithms. Therefore, SYMPOL can become the foundation for a new class of interpretable RL based on decision trees. Our implementation is available under: https://github.com/s-marton/sympol