COLT
2000
Springer
13 years 9 months ago
2000
Springer
We describe a novel family of PAC model algorithms for learning linear threshold functions. The new algorithms work by boosting a simple weak learner and exhibit complexity bounds...