Generating Explanations for Conceptual Validation of Graph Neural Networks: An Investigation of Symbolic Predicates Learned on Relevance-Ranked Sub-Graphs.
Bettina FinzelAnna SarantiAlessa AngerschmidDavid E. TaflerBastian PfeiferAndreas HolzingerPublished in: Künstliche Intell. (2022)
Keyphrases
- generating explanations
- neural network
- graph representation
- weighted graph
- graph theory
- graph matching
- directed graph
- labeled graphs
- graph structure
- graph theoretic
- graph mining
- graph model
- graph databases
- graph structures
- graph construction
- adjacency matrix
- series parallel
- graph partitioning
- graph theoretical
- connectionist systems
- graph search
- graph clustering
- graph classification
- connectionist models
- random graphs
- graph properties
- undirected graph
- graph kernels
- spanning tree
- pattern recognition
- subgraph isomorphism
- bipartite graph
- maximum common subgraph
- ranked list
- minimum spanning tree
- descending order
- neural learning
- graph isomorphism
- graph representations
- graph data
- information retrieval
- connected graphs
- connected dominating set
- web graph
- artificial neural networks
- maximum cardinality
- dense subgraphs
- dynamic graph
- graph transformation
- retrieved documents
- conceptual spaces
- edge weights
- graph layout
- real world graphs
- reachability queries
- connected components
- attributed graphs
- inexact graph matching
- structured data
- topological information
- community discovery
- directed acyclic
- directed acyclic graph
- graph embedding
- rank order
- result list
- maximum clique
- relevance feedback
- neighborhood graph
- maximal cliques
- graph patterns
- evolving graphs
- ranking algorithm
- planar graphs
- polynomial time complexity
- query graph
- small world