Gradient Derivation for Learnable Parameters in Graph Attention Networks.
Marion NeumeierAndreas TollkühnSebastian DornMichael BotschWolfgang UtschickPublished in: CoRR (2023)
Keyphrases
- network parameters
- parameter values
- edge weights
- graph representation
- steepest ascent
- dynamic networks
- graph structure
- community discovery
- parameter estimation
- connected components
- complex networks
- weighted graph
- network analysis
- random walk
- graph theoretic
- graph structures
- clustering coefficient
- graph mining
- small world
- graph model
- social networks
- visual attention
- expectation maximization
- image segmentation