Active Learning without Knowing Individual Instance Labels: A Pairwise Label Homogeneity Query Approach.
Yifan FuBin LiXingquan ZhuChengqi ZhangPublished in: IEEE Trans. Knowl. Data Eng. (2014)
Keyphrases
- active learner
- active learning
- pairwise
- unlabeled instances
- multiple annotators
- label noise
- relevance feedback
- semi supervised
- active learning framework
- labeling effort
- query by committee
- labeled data
- label ranking
- unlabeled data
- query processing
- supervised learning
- training instances
- training examples
- unlabeled examples
- version space
- transfer learning
- labeled examples
- database
- learning algorithm
- label propagation
- training data
- multi class
- image labeling
- machine learning
- multi label
- loss function
- similarity measure
- pairwise constraints
- semi supervised learning
- training set
- response time
- annotation effort
- labeled instances
- object retrieval
- random sampling
- cost sensitive
- user queries
- multi label learning
- instance level
- class labels
- labeling process
- multi instance
- markov random field
- multi label classification
- decision trees
- information retrieval