Quantifying the Intrinsic Usefulness of Attributional Explanations for Graph Neural Networks with Artificial Simulatability Studies.
Jonas TeufelLuca TorresiPascal FriederichPublished in: CoRR (2023)
Keyphrases
- neural network
- pattern recognition
- genetic algorithm
- graph theory
- graph model
- artificial neural networks
- structured data
- dependency graph
- weighted graph
- graph structure
- graph matching
- random walk
- empirical studies
- directed graph
- data sets
- multilayer perceptron
- connected components
- recurrent neural networks
- bipartite graph
- bayesian networks
- case study
- real world
- graph partitioning
- activation function
- literature review
- graph theoretic
- edge weights
- graph search