图神经网络中条件方法的探索（英文版）

英文标题：An Exploration of Conditioning Methods in Graph Neural Networks中文摘要：本研究针对使用节点和边缘属性提高图神经网络性能的三种条件方法（弱条件、强条件和纯条件）进行了实证研究，并将其应用于计算化学的多项任务进行分析。英文摘要：The flexibility and effectiveness of message passing based graph neuralnetworks (GNNs) induced considerable advances in deep learning ongraph-structured data. In such approaches, GNNs recursively update noderepresentations based on their neighbors and they gain expressivity through theuse of node and edge attribute vectors. E.g., in computational ta