Hijacking Attacks against Neural Network by Analyzing Training Data.
Yunjie GeQian WangHuayang HuangQi LiCong WangChao ShenLingchen ZhaoPeipei JiangZheng FangShenyi ZhangPublished in: USENIX Security Symposium (2024)
Keyphrases
- training data
- neural network
- training process
- back propagation
- training set
- artificial neural networks
- learning algorithm
- decision trees
- data sets
- training patterns
- supervised learning
- classification accuracy
- domain knowledge
- neural nets
- countermeasures
- neural network model
- labeled data
- malicious attacks
- noisy data
- training dataset
- activation function
- watermarking scheme
- multi layer
- training instances
- security threats
- fuzzy artmap
- test set
- training examples
- fuzzy logic
- prior knowledge
- malicious users
- feed forward neural networks
- neural network is trained
- network architecture
- classification models
- feed forward
- test data
- fault diagnosis
- training samples
- semi supervised learning
- text classification