Optimizing Interpretable Decision Tree Policies for Reinforcement Learning
This addresses the need for interpretable AI policies in reinforcement learning, though it is incremental as it builds on existing decision tree and policy gradient techniques.
The paper tackles the problem of replacing complex neural network policies with interpretable decision tree policies in reinforcement learning, proposing the Decision Tree Policy Optimization (DTPO) algorithm that directly optimizes complete trees using policy gradients and shows competitive performance compared to imitation learning methods.
Reinforcement learning techniques leveraging deep learning have made tremendous progress in recent years. However, the complexity of neural networks prevents practitioners from understanding their behavior. Decision trees have gained increased attention in supervised learning for their inherent interpretability, enabling modelers to understand the exact prediction process after learning. This paper considers the problem of optimizing interpretable decision tree policies to replace neural networks in reinforcement learning settings. Previous works have relaxed the tree structure, restricted to optimizing only tree leaves, or applied imitation learning techniques to approximately copy the behavior of a neural network policy with a decision tree. We propose the Decision Tree Policy Optimization (DTPO) algorithm that directly optimizes the complete decision tree using policy gradients. Our technique uses established decision tree heuristics for regression to perform policy optimization. We empirically show that DTPO is a competitive algorithm compared to imitation learning algorithms for optimizing decision tree policies in reinforcement learning.