Statements (49)
| Predicate | Object |
|---|---|
| gptkbp:instance_of |
gptkb:neural_networks
|
| gptkbp:applies_to |
large-scale graphs
|
| gptkbp:can_be_used_in |
recommendation systems
social network analysis biological network analysis knowledge graph completion |
| gptkbp:developed_by |
gptkb:Stanford_University
|
| gptkbp:enables |
inductive learning
|
| gptkbp:has_applications_in |
traffic prediction
image classification anomaly detection community detection text classification recommendation engines link prediction financial fraud detection |
| gptkbp:has_limitations |
requires careful tuning
computationally intensive for large graphs depends on quality of node features may overfit on small datasets sensitive to noise |
| gptkbp:has_variants |
Graph SAGE with LSTM
Graph SAGE with attention Graph SAGE with pooling |
| https://www.w3.org/2000/01/rdf-schema#label |
Graph SAGE
|
| gptkbp:influenced_by |
Deep Walk
Node2 Vec |
| gptkbp:introduced_in |
gptkb:2017
|
| gptkbp:is_compared_to |
gptkb:GCN
GAT |
| gptkbp:is_evaluated_by |
Citeseer dataset
Cora dataset Pubmed dataset Reddit dataset ogbn-arxiv dataset |
| gptkbp:is_implemented_in |
gptkb:Tensor_Flow
gptkb:Py_Torch |
| gptkbp:is_notable_for |
flexibility
scalability ability to handle heterogeneous graphs ability to incorporate node features performance on unseen data |
| gptkbp:is_part_of |
graph representation learning
|
| gptkbp:supports |
dynamic graphs
|
| gptkbp:used_for |
node classification
|
| gptkbp:uses |
sampling techniques
|
| gptkbp:utilizes |
neighborhood aggregation
|
| gptkbp:bfsParent |
gptkb:neural_networks
|
| gptkbp:bfsLayer |
4
|