Simplified GCN

GPTKB entity

Statements (49)
Predicate Object
gptkbp:instanceOf graph neural network model
gptkbp:abbreviation gptkb:SGC
gptkbp:advantage lower memory usage
competitive accuracy
faster than standard GCN
gptkbp:application semi-supervised learning
graph representation learning
node classification
gptkbp:arXivID 1902.07153
gptkbp:basedOn gptkb:Graph_Convolutional_Network
gptkbp:benchmarkDataset gptkb:Cora
gptkb:Citeseer
Pubmed
gptkbp:citation gptkb:Simplifying_Graph_Convolutional_Networks
10.48550/arXiv.1902.07153
gptkbp:contribution enables faster training
improves efficiency of GCNs
reduces overfitting
removes nonlinearities and collapses weight matrices between GCN layers
gptkbp:fullName gptkb:Simplified_Graph_Convolutional_Network
https://www.w3.org/2000/01/rdf-schema#label Simplified GCN
gptkbp:input graph
adjacency matrix
node features
gptkbp:key precompute K-step feature propagation
use a single linear classifier after propagation
gptkbp:limitation cannot model complex nonlinear relationships
less expressive than deep GCNs
gptkbp:output node labels
node embeddings
gptkbp:proposedBy gptkb:Zhiwei_Steven_Wu
gptkb:Tianyi_Zhang
gptkb:Felix_Wu
gptkb:Amauri_Holanda_de_Souza_Jr.
gptkb:Khaled_S._Refaat
gptkb:Kirill_Levchenko
gptkb:Tianjian_Meng
Christopher Fifty
gptkbp:publicationYear 2019
gptkbp:publishedIn gptkb:ICML_2019
gptkbp:relatedTo gptkb:GCN
gptkb:convolutional_neural_network
gptkb:Graph_Convolutional_Network
gptkbp:repository https://github.com/Tiiiger/SGC
gptkbp:usedIn social networks
biological networks
citation networks
gptkbp:bfsParent gptkb:GCN
gptkbp:bfsLayer 6