It is double pleasure to deceive the deceiver: disturbing classifiers against adversarial attacks.
João G. ZagoEric A. AntoneloFabio L. BaldisseraRodrigo T. SaadPublished in: SMC (2020)
Keyphrases
- decision trees
- support vector
- training data
- supervised classification
- naive bayes
- linear classifiers
- multi agent
- countermeasures
- feature set
- feature selection
- class labels
- machine learning algorithms
- training samples
- classification algorithm
- watermarking scheme
- extracted features
- anomaly detection
- active learning
- chosen plaintext
- security threats
- majority voting
- ensemble classifier
- malicious users
- classification systems
- network traffic
- classification method
- svm classifier
- training examples
- intrusion detection
- training set