Partition-Based Active Learning for Graph Neural Networks
This addresses the challenge of efficiently labeling nodes in graph-based learning, which is incremental as it builds on existing active learning approaches for GNNs.
The paper tackles the problem of semi-supervised learning with Graph Neural Networks in an active learning setup by proposing GraphPart, a partition-based method that selects representative nodes to query, and it outperforms existing methods across various annotation budgets in experiments on benchmark datasets.
We study the problem of semi-supervised learning with Graph Neural Networks (GNNs) in an active learning setup. We propose GraphPart, a novel partition-based active learning approach for GNNs. GraphPart first splits the graph into disjoint partitions and then selects representative nodes within each partition to query. The proposed method is motivated by a novel analysis of the classification error under realistic smoothness assumptions over the graph and the node features. Extensive experiments on multiple benchmark datasets demonstrate that the proposed method outperforms existing active learning methods for GNNs under a wide range of annotation budget constraints. In addition, the proposed method does not introduce additional hyperparameters, which is crucial for model training, especially in the active learning setting where a labeled validation set may not be available.