Monitoring Classification Blindspots to Detect Drifts from Unlabeled Data.
Tegjyot Singh SethiMehmed M. KantardzicElaheh ArabmakkiPublished in: IRI (2016)
Keyphrases
- unlabeled data
- labeled data
- co training
- text classification
- class labels
- semi supervised learning
- semi supervised classification
- semi supervised
- supervised learning
- labeled and unlabeled data
- improve the classification accuracy
- supervised learning algorithms
- training set
- active learning
- semisupervised learning
- training data
- semi supervised learning algorithms
- decision boundary
- label information
- text categorization
- labeled instances
- machine learning
- learning algorithm
- unsupervised learning
- labeled examples
- classification accuracy
- domain adaptation
- feature vectors
- select relevant features
- feature extraction
- document classification
- data points
- transfer learning
- feature selection
- unlabeled samples
- labeled training data
- training examples
- support vector
- class distribution
- pattern classification
- decision trees
- learning tasks
- supervised and semi supervised
- high dimensional
- feature space
- multiple instance learning
- transductive learning
- training samples
- data sets
- labeled data for training
- classification algorithm