E-SAGE: Explainability-based Defense Against Backdoor Attacks on Graph Neural Networks.
Dingqiang YuanXiaohua XuLei YuTongchang HanRongchang LiMeng HanPublished in: CoRR (2024)
Keyphrases
- neural network
- countermeasures
- graph representation
- artificial neural networks
- ddos attacks
- graph structure
- back propagation
- directed graph
- graph theory
- neural nets
- intrusion detection
- pattern recognition
- graph theoretic
- recently developed
- activation function
- bipartite graph
- multi layer
- computer virus
- watermarking algorithm
- graph model
- defense mechanisms
- weighted graph
- neural network model
- random walk
- fuzzy logic
- network security
- associative memory
- multilayer perceptron
- graph partitioning
- connected components
- self organizing maps
- security threats
- learning algorithm