Research Explores Backdoor Defense in Graph Neural Networks
Recent research explores data isolation techniques to enhance the security of Graph Neural Networks against backdoor attacks, highlighting methods for identifying and mitigating compromised samples during model training.