Graph neural networks (GNNs) have shown great promise in few-shot learning. where they typically represent the entire feature of a sample as a node. However. this approach can overlook finer details within the sample. https://www.ngetikin.com/