Redundant Noisy Attributes, Attribute Errors, and Linear-Threshold Learning Using Winnow.
Nick LittlestonePublished in: COLT (1991)
Keyphrases
- linear threshold
- winnow algorithm
- learning algorithm
- boolean functions
- learning problems
- sampling algorithm
- perceptron algorithm
- supervised learning
- attribute values
- active learning
- sample complexity
- theoretical analysis
- decision trees
- irrelevant attributes
- generative model
- concept classes
- mistake bound
- feature selection