Agreeing to disagree: active learning with noisy labels without crowdsourcing.
Mohamed-Rafik BougueliaSlawomir NowaczykK. C. SantoshAntanas VerikasPublished in: Int. J. Mach. Learn. Cybern. (2018)
Keyphrases
- active learning
- label noise
- crowd sourced
- annotation effort
- training examples
- training set
- unlabeled examples
- unlabeled instances
- learning strategies
- active learning framework
- selective sampling
- labeling effort
- supervised learning
- labeled data
- semi supervised
- class labels
- machine learning
- pairwise
- experimental design
- labeled instances
- sample selection
- learning algorithm
- labeled examples
- random sampling
- cost sensitive
- noisy data
- active learner
- semi supervised learning
- unlabeled data
- music emotion classification
- pool based active learning
- training data
- relevance feedback
- active learning strategies
- noisy environments
- neural network
- human computation
- label propagation