Graph Neural Networks (GNNs) are vulnerable to backdoor attacks, where triggers inserted into original graphs cause adversary-determined predictions. Backdoor attacks on GNNs, typically focusing on ...