MCDAL: Maximum Classifier Discrepancy for Active Learning.
Jae-Won ChoDong-Jin KimYunjae JungIn So KweonPublished in: CoRR (2021)
Keyphrases
- active learning
- training set
- training examples
- label noise
- learning algorithm
- training data
- learning process
- support vector machine
- query by committee
- random sampling
- supervised learning
- feature space
- classification method
- decision trees
- classification scheme
- classification algorithm
- multiple classifiers
- svm classifier
- classifier systems
- cost sensitive
- learning strategies
- cost sensitive learning
- semi supervised
- linear classifiers
- binary classifiers
- labeled instances
- representative samples
- unlabeled instances
- active learning framework
- active learner
- transfer learning
- training samples
- feature set
- multi class
- support vector
- feature selection
- machine learning
- classification process
- experimental design
- training instances
- relevance feedback
- active learning strategies
- labeling effort
- feature extraction
- data sets