Verifying Attention Robustness of Deep Neural Networks against Semantic Perturbations.
Satoshi MunakataCaterina UrbanHaruki YokoyamaKoji YamamotoKazuki MunakataPublished in: APSEC (2022)
Keyphrases
- neural network
- genetic algorithm
- pattern recognition
- semantic search
- neural network model
- semantic information
- natural language
- artificial neural networks
- high robustness
- semantic web
- semantic annotation
- visual attention
- high level
- semantic representation
- semantic knowledge
- network architecture
- recurrent neural networks
- semantically equivalent
- semantic level
- semantic similarity
- fuzzy logic
- semantic features
- learning algorithm
- semantic content
- multilayer perceptron
- semantically meaningful
- semantic interpretation
- fault diagnosis
- sufficient conditions