M-Walk: Learning to Walk over Graphs using Monte Carlo Tree Search
This addresses the challenge of sparse rewards in graph-walking tasks for applications like knowledge base completion, though it is incremental as it builds on existing RL and MCTS techniques.
The paper tackles the problem of learning to walk over graphs towards a target node, such as in knowledge base completion, by developing M-Walk, an agent combining a deep RNN with Monte Carlo Tree Search to handle sparse rewards. It shows that M-Walk learns better policies than other RL-based methods and outperforms traditional KBC baselines in experiments on graph-walking benchmarks.
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.