Researchers have developed a new method for explaining how graph neural networks make predictions. The method, called NICE, uses a technique called Noise Corruption to perturb messages in the network without causing a 'scale drift' that can lead to unreliable explanations. Scale drift occurs when the network's predictions are influenced by the way it handles edge-induced messages. NICE learns a boundary between correct and incorrect predictions under uncertainty, allowing for more accurate explanations. The authors claim their method outperforms existing approaches on multiple benchmarks. --- Why it matters: This matters because graph neural networks are widely used in applications like social network analysis and traffic forecasting, where understanding how the model makes predictions is crucial. NICE's ability to provide more accurate explanations can improve trust in these models and enable better decision-making. Source: https://arxiv.org/abs/2608.16038