Training-free Graph Neural Networks and the Power of Labels as Features
This addresses the need for efficient and powerful graph learning methods, offering a novel approach that reduces computational costs, though it is incremental in building on existing GNN frameworks.
The paper tackles transductive node classification by proposing training-free graph neural networks (TFGNNs) that use labels as features, showing they outperform existing GNNs without training and converge faster with optional training.
We propose training-free graph neural networks (TFGNNs), which can be used without training and can also be improved with optional training, for transductive node classification. We first advocate labels as features (LaF), which is an admissible but not explored technique. We show that LaF provably enhances the expressive power of graph neural networks. We design TFGNNs based on this analysis. In the experiments, we confirm that TFGNNs outperform existing GNNs in the training-free setting and converge with much fewer training iterations than traditional GNNs.