Interpretation of Neural Networks is Susceptible to Universal Adversarial Perturbations.
Haniyeh Ehsani OskouieFarzan FarniaPublished in: CoRR (2022)
Keyphrases
- neural network
- artificial neural networks
- image interpretation
- pattern recognition
- back propagation
- self organizing maps
- high level
- multilayer perceptron
- fault diagnosis
- semantic interpretation
- multi valued
- activation function
- fuzzy systems
- neural nets
- associative memory
- neural network model
- learning algorithm
- genetic algorithm
- database
- neural network is trained
- kolmogorov complexity
- neural learning
- decision trees
- data sets