ImpNet: Imperceptible and blackbox-undetectable backdoors in compiled neural networks.
Tim CliffordIlia ShumailovYiren ZhaoRoss J. AndersonRobert MullinsPublished in: CoRR (2022)
Keyphrases
- neural network
- pattern recognition
- artificial neural networks
- genetic algorithm
- fuzzy logic
- fuzzy systems
- multi layer
- machine learning
- watermarking scheme
- neural nets
- feed forward
- back propagation
- network architecture
- data sets
- self organizing maps
- orders of magnitude
- expert systems
- computational complexity
- multiscale
- artificial intelligence
- activation function
- hidden structure