Matthias Seeger
We emphasize the need for input-dependent regularization in the context of conditional density models (also: discriminative models) like Gaussian process predictors. This can be achieved by a simple modification of the standard Bayesian data generation mod ...
2000