Graph Attention Network

Intermediate

GNN using attention to weight neighbor contributions dynamically.

Full Definition

GNN using attention to weight neighbor contributions dynamically.

Keywords

attention weights

Domains

Model Architectures

Related Terms

Heterogeneous Graph related to

Graphs containing multiple node or edge types with different semantics.

 Knowledge Graph related to

Structured graph encoding facts as entity–relation–entity triples.

 Energy-Based Model related to

Models that define an energy landscape rather than explicit probabilities.

 Boltzmann Machine related to

Probabilistic energy-based neural network with hidden variables.

 Graph Neural Network related to

Neural networks that operate on graph-structured data by propagating information along edges.

 Message Passing Neural Network related to

GNN framework where nodes iteratively exchange and aggregate messages from neighbors.

 Graph Convolution related to

Extension of convolution to graph domains using adjacency structure.

Concept Map

See how Graph Attention Network connects to other concepts.

Open Knowledge Graph