TreeQN and ATreeC: Differentiable Tree-Structured Models for Deep Reinforcement Learning

TL;DR

This paper introduces TreeQN and ATreeC, differentiable tree-structured models for deep reinforcement learning that improve planning and value estimation in complex environments by learning transition models end-to-end.

Contribution

The paper presents TreeQN and ATreeC, novel differentiable tree-structured models that integrate learned transition models into deep RL, enabling better planning and policy learning.

Findings

TreeQN and ATreeC outperform baseline methods on a box-pushing task.

They also outperform n-step DQN and value prediction networks on Atari games.

Ablation studies show the importance of auxiliary losses for learning transition models.

Abstract

Combining deep model-free reinforcement learning with on-line planning is a promising approach to building on the successes of deep RL. On-line planning with look-ahead trees has proven successful in environments where transition models are known a priori. However, in complex environments where transition models need to be learned from data, the deficiencies of learned models have limited their utility for planning. To address these challenges, we propose TreeQN, a differentiable, recursive, tree-structured model that serves as a drop-in replacement for any value function network in deep RL with discrete actions. TreeQN dynamically constructs a tree by recursively applying a transition model in a learned abstract state space and then aggregating predicted rewards and state-values using a tree backup to estimate Q-values. We also propose ATreeC, an actor-critic variant that augments TreeQN with a softmax layer to form a stochastic policy network. Both approaches are trained end-to-end, such that the learned model is optimised for its actual use in the tree. We show that TreeQN and ATreeC outperform n-step DQN and A2C on a box-pushing task, as well as n-step DQN and value prediction networks (Oh et al. 2017) on multiple Atari games. Furthermore, we present ablation studies that demonstrate the effect of different auxiliary losses on learning transition models.