Non-greedy active learning for text categorization using convex transductive experimental design.
Kai YuShenghuo ZhuWei XuYihong GongPublished in: SIGIR (2008)
Keyphrases
- text categorization
- experimental design
- active learning
- unlabeled data
- feature selection
- semi supervised learning
- text classification
- labeled data
- semi supervised
- uncertainty sampling
- ensemble pruning
- multi label
- supervised learning
- machine learning
- information gain
- training set
- transfer learning
- learning algorithm
- class imbalance
- naive bayes
- support vector
- learning strategies
- relevance feedback
- empirical studies
- knn
- decision trees
- model selection
- training examples
- k nearest neighbor
- class distribution
- learning process
- generalization error
- neural network
- unsupervised learning
- data mining
- e learning
- reinforcement learning
- cost sensitive
- classification accuracy
- sample size
- high dimensional data
- support vector machine
- data sets