用于支持论文《Relational inductive biases, deep learning, and graph networks》。
A graph network takes a graph as input and returns a graph as output. The input graph has edge- (E ), node- (V ), and global-level (u) attributes. The output graph has the same structure, but updated attributes. Graph networks are part of the broader family of "graph neural networks" (Scarselli et al., 2009).
讲直白一些,就是用神经网络处理图,输入是图,输出也是图。以前都是处理向量(Vector),所以NLP中需要做Word2Vec后才能运用深度学习的处理结果。作者们认为 Graph2Graph 是让神经网络具备推理(Reason)能力的一个关键步骤。
【CNN已老,GNN来了】DeepMind、谷歌大脑、MIT等27位作者重磅论文,图网络让深度学习也能因果推理