128
Voted
ESANN
2001
15 years 4 months ago
2001
We develop improved risk bounds for function estimation with models such as single hidden layer neural nets, using a penalized least squares criterion to select the size of the mod...