The relative value of labeled and unlabeled samples in pattern recognition with an unknown mixing parameter.
Vittorio CastelliThomas M. CoverPublished in: IEEE Trans. Inf. Theory (1996)
Keyphrases
- unlabeled samples
- pattern recognition
- semi supervised
- semi supervised learning
- incremental learning
- labeled samples
- labeled data
- data distribution
- training samples
- semi supervised classification
- em algorithm
- labeled training data
- unlabeled data
- mixture model
- label propagation
- label information
- labeled and unlabeled data
- supervised learning
- error reduction
- dimensionality reduction
- discriminative information
- image processing
- high confidence
- training data
- active learning
- multiple features
- support vector machine svm
- pattern classification
- machine learning
- association rules
- computer vision
- generative model
- labeled examples
- co training
- text mining
- classification accuracy
- pairwise
- data streams
- data sets