Policy Tree: Adaptive Representation for Policy Gradient


  • Ujjwal Das Gupta University of Alberta
  • Erik Talvitie Franklin and Marshall College
  • Michael Bowling University of Alberta




Reinforcement Learning, Policy Gradient, Decision Trees, Representation Learning


Much of the focus on finding good representations in reinforcement learning has been on learning complex non-linear predictors of value. Policy gradient algorithms, which directly represent the policy, often need fewer parameters to learn good policies. However, they typically employ a fixed parametric representation that may not be sufficient for complex domains. This paper introduces the Policy Tree algorithm, which can learn an adaptive representation of policy in the form of a decision tree over different instantiations of a base policy. Policy gradient is used both to optimize the parameters and to grow the tree by choosing splits that enable the maximum local increase in the expected return of the policy. Experiments show that this algorithm can choose genuinely helpful splits and significantly improve upon the commonly used linear Gibbs softmax policy, which we choose as our base policy.




How to Cite

Das Gupta, U., Talvitie, E., & Bowling, M. (2015). Policy Tree: Adaptive Representation for Policy Gradient. Proceedings of the AAAI Conference on Artificial Intelligence, 29(1). https://doi.org/10.1609/aaai.v29i1.9613



Main Track: Novel Machine Learning Algorithms