GANI: Global Attacks on Graph Neural Networks via Imperceptible Node Injections.
Junyuan FangHaixian WenJiajing WuQi XuanZibin ZhengChi K. TsePublished in: CoRR (2022)
Keyphrases
- neural network
- graph structure
- directed graph
- watermarking scheme
- finding the shortest path
- watermarking algorithm
- random walk
- connected components
- genetic algorithm
- watermarking technique
- countermeasures
- undirected graph
- graph theory
- weighted graph
- fuzzy logic
- edge weights
- graph partitioning
- nodes of a graph
- pattern recognition
- graph model
- structured data
- back propagation
- digital images
- spatial domain
- graph representation
- data hiding
- spanning tree
- security mechanisms
- path length
- root node
- self organizing maps
- recurrent neural networks