Timezone: »
We investigate the representation power of graph neural networks in the semi-supervised node classification task under heterophily or low homophily, i.e., in networks where connected nodes may have different class labels and dissimilar features. Many popular GNNs fail to generalize to this setting, and are even outperformed by models that ignore the graph structure (e.g., multilayer perceptrons). Motivated by this limitation, we identify a set of key designs—ego- and neighbor-embedding separation, higher-order neighborhoods, and combination of intermediate representations—that boost learning from the graph structure under heterophily. We combine them into a graph neural network, H2GCN, which we use as the base method to empirically evaluate the effectiveness of the identified designs. Going beyond the traditional benchmarks with strong homophily, our empirical analysis shows that the identified designs increase the accuracy of GNNs by up to 40% and 27% over models without them on synthetic and real networks with heterophily, respectively, and yield competitive performance under homophily.
Author Information
Jiong Zhu (University of Michigan)
Yujun Yan (University of Michigan)
Lingxiao Zhao (Carnegie Mellon University)
Mark Heimann (Lawrence Livermore National Laboratory)
Leman Akoglu (CMU)
Danai Koutra (U Michigan)
More from the Same Authors
-
2021 : Two Sides of the Same Coin: Heterophily and Oversmoothing in Graph Convolutional Neural Networks »
Yujun Yan · Milad Hashemi · Kevin Swersky · Yaoqing Yang · Danai Koutra -
2021 : A Graph Perspective on Neural Network Dynamics »
Fatemeh Vahedian · Ruiyu Li · Puja Trivedi · Di Jin · Danai Koutra -
2022 : A Closer Look at Model Adaptation using Feature Distortion and Simplicity Bias »
Puja Trivedi · Danai Koutra · Jayaraman Thiagarajan -
2022 : Modeling Hierarchical Topological Structure in Scientific Images with Graph Neural Networks »
Samuel Leventhal · Attila Gyulassy · Valerio Pascucci · Mark Heimann -
2022 Poster: Hyperparameter Sensitivity in Deep Outlier Detection: Analysis and a Scalable Hyper-Ensemble Solution »
Xueying Ding · Lingxiao Zhao · Leman Akoglu -
2022 Poster: Analyzing Data-Centric Properties for Graph Contrastive Learning »
Puja Trivedi · Ekdeep S Lubana · Mark Heimann · Danai Koutra · Jayaraman Thiagarajan -
2022 Poster: A Practical, Progressively-Expressive GNN »
Lingxiao Zhao · Neil Shah · Leman Akoglu -
2022 Poster: Dual-discriminative Graph Neural Network for Imbalanced Graph-level Anomaly Detection »
GE ZHANG · Zhenyu Yang · Jia Wu · Jian Yang · Shan Xue · Hao Peng · Jianlin Su · Chuan Zhou · Quan Z. Sheng · Leman Akoglu · Charu Aggarwal -
2021 Poster: Automatic Unsupervised Outlier Model Selection »
Yue Zhao · Ryan Rossi · Leman Akoglu -
2020 Poster: Neural Execution Engines: Learning to Execute Subroutines »
Yujun Yan · Kevin Swersky · Danai Koutra · Parthasarathy Ranganathan · Milad Hashemi -
2019 Poster: Statistical Analysis of Nearest Neighbor Methods for Anomaly Detection »
Xiaoyi Gu · Leman Akoglu · Alessandro Rinaldo