An Investigation on the Fragility of Graph Neural Networks: The Impact of Node Feature Modification on Graph Classification Accuracy.
Chengen WangYan ZhouKangkook JeeMurat KantarciogluPublished in: TPS-ISA (2023)
Keyphrases
- neural network
- graph structure
- directed graph
- classification accuracy
- graph representation
- graph model
- graph theory
- graph theoretic
- feature vectors
- artificial neural networks
- image features
- edge weights
- random walk
- pattern recognition
- graph construction
- connected components
- neural network model
- data sets
- machine learning
- undirected graph
- graph data
- finding the shortest path
- nodes of a graph
- weighted graph
- self organizing maps
- structured data
- feature set
- knn
- similarity measure
- image segmentation
- feature selection