Boosting Training for PDF Malware Classifier via Active Learning.
Xinxin WangYuanzhang LiQuanxin ZhangXiaohui KuangPublished in: CSS (2) (2019)
Keyphrases
- active learning
- training examples
- training set
- partially labeled data
- boosted classifiers
- weak classifiers
- learning algorithm
- ground truth labels
- query by committee
- training samples
- batch mode
- training process
- supervised learning
- early stopping
- rare classes
- training error
- discriminative classifiers
- weak learners
- machine learning
- training data
- labeled instances
- active learner
- adaboost algorithm
- feature selection
- label noise
- ensemble learning
- support vector machine
- decision stumps
- semi supervised
- cost sensitive
- decision trees
- face detection
- probability density function
- annotation effort
- object detectors
- ensemble classifier
- sample selection
- object detection
- test set
- text classifiers
- random sampling
- feature space
- support vector
- co training
- selective sampling
- unlabeled data
- training dataset
- base classifiers
- labeling effort
- classification algorithm
- multiclass classification
- learning process
- improving classification accuracy
- naive bayes
- multiple classifier systems
- class labels
- transfer learning
- training instances
- boosting algorithms
- combining multiple
- random selection
- multi layer perceptron