Optimal Oracle Inequality for Aggregation of Classifiers Under Low Noise Condition.
Guillaume LecuéPublished in: COLT (2006)
Keyphrases
- low signal to noise ratio
- training data
- high noise
- supervised classification
- sufficient conditions
- test set
- missing data
- linear classifiers
- noise level
- signal to noise ratio
- noisy data
- training set
- optimal solution
- decision trees
- feature selection
- data sets
- classification algorithm
- training examples
- training samples
- estimation error
- noise free
- random noise
- neural network