Towards Link Characterization From Content: Recovering Distributions From Classifier Output.
John GrothendieckAllen L. GorinPublished in: IEEE Trans. Speech Audio Process. (2008)
Keyphrases
- input data
- training data
- metadata
- linear classifiers
- learning algorithm
- data sets
- multimedia
- classification rate
- classification algorithm
- random variables
- probability distribution
- highly skewed
- classification process
- support vector machine
- active learning
- feature space
- support vector
- feature extraction
- training samples
- web documents
- classification method
- multimedia content
- classification scheme
- digital libraries
- power law
- classifier systems
- training and test data
- feature selection