TY - GEN
T1 - A Unified View on Graph Neural Networks as Graph Signal Denoising
AU - Ma, Yao
AU - Liu, Xiaorui
AU - Zhao, Tong
AU - Liu, Yozen
AU - Tang, Jiliang
AU - Shah, Neil
N1 - Publisher Copyright:
© 2021 ACM.
PY - 2021/10/26
Y1 - 2021/10/26
N2 - Graph Neural Networks (GNNs) have risen to prominence in learning representations for graph structured data. A single GNN layer typically consists of a feature transformation and a feature aggregation operation. The former normally uses feed-forward networks to transform features, while the latter aggregates the transformed features over the graph. Numerous recent works have proposed GNN models with different designs in the aggregation operation. In this work, we establish mathematically that the aggregation processes in a group of representative GNN models including GCN, GAT, PPNP, and APPNP can be regarded as (approximately) solving a graph denoising problem with a smoothness assumption. Such a unified view across GNNs not only provides a new perspective to understand a variety of aggregation operations but also enables us to develop a unified graph neural network framework UGNN. To demonstrate its promising potential, we instantiate a novel GNN model, ADA-UGNN, derived from UGNN, to handle graphs with adaptive smoothness across nodes. Comprehensive experiments show the effectiveness of ADA-UGNN.
AB - Graph Neural Networks (GNNs) have risen to prominence in learning representations for graph structured data. A single GNN layer typically consists of a feature transformation and a feature aggregation operation. The former normally uses feed-forward networks to transform features, while the latter aggregates the transformed features over the graph. Numerous recent works have proposed GNN models with different designs in the aggregation operation. In this work, we establish mathematically that the aggregation processes in a group of representative GNN models including GCN, GAT, PPNP, and APPNP can be regarded as (approximately) solving a graph denoising problem with a smoothness assumption. Such a unified view across GNNs not only provides a new perspective to understand a variety of aggregation operations but also enables us to develop a unified graph neural network framework UGNN. To demonstrate its promising potential, we instantiate a novel GNN model, ADA-UGNN, derived from UGNN, to handle graphs with adaptive smoothness across nodes. Comprehensive experiments show the effectiveness of ADA-UGNN.
KW - graph neural networks
KW - graph signal denoising
KW - semi-supervised classification
UR - http://www.scopus.com/inward/record.url?scp=85119201151&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85119201151&partnerID=8YFLogxK
U2 - 10.1145/3459637.3482225
DO - 10.1145/3459637.3482225
M3 - Conference contribution
AN - SCOPUS:85119201151
T3 - International Conference on Information and Knowledge Management, Proceedings
SP - 1202
EP - 1211
BT - CIKM 2021 - Proceedings of the 30th ACM International Conference on Information and Knowledge Management
PB - Association for Computing Machinery
T2 - 30th ACM International Conference on Information and Knowledge Management, CIKM 2021
Y2 - 1 November 2021 through 5 November 2021
ER -