Graph Attention Networks

GPTKB entity

Statements (55)
Predicate Object
gptkbp:instanceOf gptkb:convolutional_neural_network
gptkbp:abbreviation gptkb:GAT
gptkbp:activatedBy gptkb:ELU
gptkbp:application graph classification
link prediction
node classification
gptkbp:architecture multi-head attention
layer stacking
gptkbp:attentionMechanism learns weights for neighbors
gptkbp:benchmarkDatasets gptkb:Cora
gptkb:Citeseer
Pubmed
gptkbp:category gptkb:artificial_intelligence
graph representation learning
gptkbp:citation over 10,000 (as of 2024)
gptkbp:codeAvailable https://github.com/PetarV-/GAT
gptkbp:contribution attention mechanism for graph-structured data
handles graphs with varying neighborhood sizes
improved performance on node classification
node-level attention
gptkbp:field gptkb:machine_learning
graph neural networks
gptkbp:handles inductive learning
transductive learning
https://www.w3.org/2000/01/rdf-schema#label Graph Attention Networks
gptkbp:input adjacency matrix
node features
gptkbp:inspiredBy attention mechanism
gptkbp:introduced gptkb:Yoshua_Bengio
gptkb:Adriana_Romero
gptkb:Arantxa_Casanova
gptkb:Guillem_Cucurull
gptkb:Petar_Veličković
gptkb:Pietro_Liò
gptkbp:introducedIn 2017
gptkbp:lossFunction cross-entropy
gptkbp:mainPaperTitle gptkb:Graph_Attention_Networks
gptkbp:mainPaperURL https://arxiv.org/abs/1710.10903
gptkbp:openSource gptkb:TensorFlow_GNN
gptkb:PyTorch_Geometric
DGL
gptkbp:output node labels
node embeddings
gptkbp:platform deep learning
gptkbp:publishedIn gptkb:ICLR_2018
gptkbp:relatedTo gptkb:GraphSAGE
gptkb:Graph_Convolutional_Networks
gptkbp:type self-attention
gptkbp:usedIn biological network analysis
recommendation systems
social network analysis
chemoinformatics
gptkbp:bfsParent gptkb:Nvidia_cuGraph
gptkb:Graph_Attention_Network
gptkbp:bfsLayer 6