Training-Free Graph Neural Networks With Labels As Features
TFGNNs outperform existing GNNs without training & converge faster with optional training, using "labels as features" technique.
This is a Plain English Papers summary of a research paper called Training-free Graph Neural Networks and the Power of Labels as Features. If you like these kinds of analysis, you should subscribe to the AImodels.fyi newsletter or follow me on Twitter. Overview The paper proposes "training-free graph neural networks" (TFGNNs) that can be used without training and can also be improved with optional training for transductive node classification. It introduces "labels as features" (LaF), an admissible but unexplored technique, and shows that LaF can enhance the expressive power of graph...