A Data Augmentation-based Defense Method Against Adversarial Attacks in Neural Networks.
Yi ZengHan QiuGérard MemmiMeikang QiuPublished in: CoRR (2020)
Keyphrases
- synthetic data
- input data
- neural network
- prior knowledge
- test data
- data sets
- data collection
- prior information
- training data
- detection method
- information loss
- noisy data
- significant improvement
- statistical methods
- missing data
- computer systems
- missing values
- genetic algorithm
- knowledge discovery
- data sources
- data analysis
- multi agent
- objective function
- training samples
- high dimensional data
- image data
- data points
- probabilistic model
- original data
- preprocessing
- spectral clustering