Domain Knowledge Alleviates Adversarial Attacks in Multi-Label Classifiers.
Stefano MelacciGabriele CiravegnaAngelo SotgiuAmbra DemontisBattista BiggioMarco GoriFabio RoliPublished in: IEEE Trans. Pattern Anal. Mach. Intell. (2022)
Keyphrases
- multi label
- multi label classification
- domain knowledge
- class labels
- text categorization
- classifier training
- image classification
- training data
- binary classification
- image annotation
- graph cuts
- binary classifiers
- multi label learning
- multi instance
- text classification
- multi label image annotation
- meta level
- classification algorithm
- multi class
- max margin
- support vector
- feature selection
- naive bayes
- training set
- learning tasks
- decision trees
- semi supervised learning
- label assignment
- neural network
- machine learning algorithms
- hierarchical text categorization
- protein function prediction
- semi supervised
- multiclass classification
- svm classifier
- feature set
- image features
- prior knowledge
- object recognition
- automatic text classification
- computer vision