Hierarchical Reinforcement Learning for Zero-shot Generalization with Subtask Dependencies

TL;DR

This paper introduces a hierarchical reinforcement learning approach that enables agents to generalize to unseen environments by reasoning over subtask dependencies using a neural subtask graph solver, achieving efficient and effective performance.

Contribution

The paper presents a novel neural subtask graph solver with a graph reward propagation method for training, enabling zero-shot generalization in hierarchical RL with subtask dependencies.

Findings

The agent can perform complex reasoning to execute subtask graphs.

The method generalizes well to unseen subtask graphs.

Our approach outperforms Monte-Carlo tree search in efficiency.

Abstract

We introduce a new RL problem where the agent is required to generalize to a previously-unseen environment characterized by a subtask graph which describes a set of subtasks and their dependencies. Unlike existing hierarchical multitask RL approaches that explicitly describe what the agent should do at a high level, our problem only describes properties of subtasks and relationships among them, which requires the agent to perform complex reasoning to find the optimal subtask to execute. To solve this problem, we propose a neural subtask graph solver (NSGS) which encodes the subtask graph using a recursive neural network embedding. To overcome the difficulty of training, we propose a novel non-parametric gradient-based policy, graph reward propagation, to pre-train our NSGS agent and further finetune it through actor-critic method. The experimental results on two 2D visual domains show that our agent can perform complex reasoning to find a near-optimal way of executing the subtask graph and generalize well to the unseen subtask graphs. In addition, we compare our agent with a Monte-Carlo tree search (MCTS) method showing that our method is much more efficient than MCTS, and the performance of NSGS can be further improved by combining it with MCTS.