133
click to vote
NIPS
2008
15 years 4 months ago
2008
This work characterizes the generalization ability of algorithms whose predictions are linear in the input vector. To this end, we provide sharp bounds for Rademacher and Gaussian...