A=MULTI_RR(H) returns an object initialized with hyperparameters H.
Performs ridge regression independently for each column we wish
to learn, which only means inverting a single (kernel) matrix
** Hyperparameters, and their defaults
** ridge=1e-13; -- a ridge on the kernel
child=kernel; -- the kernel is stored as a member called "child"
use_kernels=1; -- if this is set to 0 a linear ridge regression or
an empirical kernel map (useful for reduced sets
of centers)
indices = [] -- indices of a reduced set of centers to be used
for learning. ([] means use all training set)
use_b=1 -- find a threshold, otherwise fix to 0
** Model
** alpha -- the weights
Xsv -- centers
** Methods:
** train, test, get_w
loo : calculate leave one out predictions (with empirical kernel, or linear)