Graph Policy Network for Transferable Active Learning on Graphs

TL;DR

This paper introduces a GNN-based policy network trained with reinforcement learning to efficiently select nodes for labeling, enabling transferable active learning strategies across various graph datasets.

Contribution

It proposes a novel transferable active learning method for GNNs using a reinforcement learning trained policy network that generalizes across different graphs.

Findings

The learned policy improves active learning efficiency on multiple datasets.

Transferability of the policy across graphs in the same and different domains is effective.

Experimental results demonstrate significant reduction in labeling costs.

Abstract

Graph neural networks (GNNs) have been attracting increasing popularity due to their simplicity and effectiveness in a variety of fields. However, a large number of labeled data is generally required to train these networks, which could be very expensive to obtain in some domains. In this paper, we study active learning for GNNs, i.e., how to efficiently label the nodes on a graph to reduce the annotation cost of training GNNs. We formulate the problem as a sequential decision process on graphs and train a GNN-based policy network with reinforcement learning to learn the optimal query strategy. By jointly training on several source graphs with full labels, we learn a transferable active learning policy which can directly generalize to unlabeled target graphs. Experimental results on multiple datasets from different domains prove the effectiveness of the learned policy in promoting active learning performance in both settings of transferring between graphs in the same domain and across different domains.