Unlabeled patterns to tighten Rademacher complexity error bounds for kernel classifiers.
Davide AnguitaAlessandro GhioLuca OnetoSandro RidellaPublished in: Pattern Recognit. Lett. (2014)
Keyphrases
- error bounds
- rademacher complexity
- kernel classifiers
- theoretical analysis
- worst case
- active learning
- unsupervised learning
- semi supervised learning
- learning algorithm
- data points
- training set
- feature space
- support vector
- training data
- machine learning
- upper bound
- supervised learning
- special case
- labeled data
- loss function