M-Walk: Learning to Walk over Graphs using Monte Carlo Tree Search

TL;DR

This paper introduces M-Walk, a novel reinforcement learning approach combining neural networks and Monte Carlo Tree Search to improve graph-walking tasks like knowledge base completion, effectively handling sparse rewards and outperforming existing methods.

Contribution

The paper presents M-Walk, a new RL algorithm integrating MCTS and neural networks for efficient graph traversal with sparse rewards, outperforming prior RL and baseline methods.

Findings

M-Walk outperforms other RL-based methods on graph-walking benchmarks.

M-Walk surpasses traditional knowledge base completion baselines.

The combined MCTS and neural policy improves reward collection during training.

Abstract

Learning to walk over a graph towards a target node for a given query and a source node is an important problem in applications such as knowledge base completion (KBC). It can be formulated as a reinforcement learning (RL) problem with a known state transition model. To overcome the challenge of sparse rewards, we develop a graph-walking agent called M-Walk, which consists of a deep recurrent neural network (RNN) and Monte Carlo Tree Search (MCTS). The RNN encodes the state (i.e., history of the walked path) and maps it separately to a policy and Q-values. In order to effectively train the agent from sparse rewards, we combine MCTS with the neural policy to generate trajectories yielding more positive rewards. From these trajectories, the network is improved in an off-policy manner using Q-learning, which modifies the RNN policy via parameter sharing. Our proposed RL algorithm repeatedly applies this policy-improvement step to learn the model. At test time, MCTS is combined with the neural policy to predict the target node. Experimental results on several graph-walking benchmarks show that M-Walk is able to learn better policies than other RL-based methods, which are mainly based on policy gradients. M-Walk also outperforms traditional KBC baselines.