3. GNNs with Heterophily 通过消息传递机制的视角,现有异染的卫星系统可以分为两组:(1) Non-local neighbor extension methods; (2) GNN architecture refinement methods。这两个类别相应地解决了两个关键问题: (1)如何发现合适的邻居; (2)如何利用发现的邻居的信息。 3. GNNs with Heterophily 通过消息传递...
Graph Neural Networks with Heterophily.Jiong ZhuRyan A. RossiAnup B. RaoTung MaiNedim LipkaNesreen K. AhmedDanai KoutraNational Conference on Artificial Intelligence
传统GNN是基于某种‘同配假设’的,也就是GNNs are built on the homophily assumption : connected nodes tend to share similar attributes with each other。(注意这里同配和上面我的理解就不同) 作者认为,传统的同配性度量是 linear feature-independent graph-label consistency,也就是frature无关的,只基于图...
图神经网络(GNN)通过使用基于关系归纳偏差(同质性假设)的图结构来扩展神经网络(Neural Network,NN)。虽然人们普遍认为GNN在实际任务中优于NN,但最近的工作发现了一组非平凡的数据集(异质图数据集),与NN相比,GNN的性能并不令人满意。异质性被认为是GNN性能不好的主要原因,并提出了许多工作来解决这一问题。在本文中...
Graph Neural Networks (GNNs) have proven to be useful for many different practical applications. However, most existing GNN models have an implicit assumption of homophily among the nodes connected in the graph, and therefore have largely overlooked the important setting of heterophily. In this wor...
Revisiting heterophily for graph neural networks. NIPS, 2022.概介绍了一种新的 graph homophily metrics.符号说明G=(V,E,A)G=(V,E,A), graph; |V|=N|V|=N; A∈RN×NA∈RN×N, adjacency matrix; Ni={j:eij∈E}Ni={j:eij∈E}, neighborhood set; X∈RN×FX∈RN×F, feature matrix; Z...
Graph Neural Networks (GNNs) have achieved remarkable success in diverse real-world applications. Traditional GNNs are designed based on homophily, which leads to poor performance under heterophily scenarios. Most current solutions deal with heterophily mainly by modeling the heterophily edges as data...
Graph neural networks with heterophily Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35 (2021), pp. 11168-11176 CrossrefGoogle Scholar [12] C. Cai, Y. Wang A note on over-smoothing for graph neural networks arXiv preprint arXiv:2006.13318 (2020) Google Scholar [13] ...
注意: heterophily 和常说的 Heterogeneity (异构图) 有点像, 后者描述的是一个图的结点的 type 不一致 (≥2≥2), 比如一个图中存在 (users, watch, movies) 的关系. 前者描述了 group 的一个性质, 显然当 h→1h→1 的时候, 不同的 groups 之间就很少产生联系, 这个时候进行 node 的分类是比较容易...
1. Understanding Heterophily for Graph Neural Networks 2. Feature Distribution on Graph Topology Mediates the Effect of Graph Convolution: Homophily Perspective 3. Graph Neural Networks Use Graphs When They Shouldn't 4. Sign is Not a Remedy: Multiset-to-Multiset Message Passing for Learning on He...