A Principled Approach to Analyze Expressiveness and Accuracy of Graph Neural Networks.
Asma AtamnaNataliya SokolovskaJean-Claude CrivelloPublished in: IDA (2020)
Keyphrases
- neural network
- high accuracy
- graph theory
- pattern recognition
- error rate
- expressive power
- fuzzy logic
- prediction accuracy
- computational cost
- random walk
- recurrent neural networks
- highly accurate
- neural nets
- graph based algorithm
- activation function
- graph structure
- feed forward
- directed graph
- self organizing maps
- np complete
- artificial neural networks
- computational complexity
- decision trees