himalaya.kernel_ridge.solve_weighted_kernel_ridge_conjugate_gradient

himalaya.kernel_ridge.solve_weighted_kernel_ridge_conjugate_gradient(Ks, Y, deltas, alpha=1.0, fit_intercept=False, initial_dual_weights=None, max_iter=100, tol=0.0001, n_targets_batch=None, random_state=None)

Solve weighted kernel ridge regression using conjugate gradient.

Solve the kernel ridge regression:

w* = argmin_w ||K @ w - Y||^2 + alpha (w.T @ K @ w)

where the kernel K is a weighted sum of multiple kernels:

K = sum_i exp(deltas[i]) Ks[i]

Parameters

Ksarray of shape (n_kernels, n_samples, n_samples)

Input kernels.

Ytorch.Tensor of shape (n_samples, n_targets)

Target data.

deltasarray of shape (n_kernels, ) or (n_kernels, n_targets)

Kernel weights.

alphafloat, or array of shape (n_targets, )

Regularization parameter.

fit_interceptboolean

Whether to fit an intercept. If False, Ks should be centered (see KernelCenterer), and Y must be zero-mean over samples.

initial_dual_weightsarray of shape (n_samples, n_targets)

Initial kernel ridge coefficients.

max_iterint

Maximum number of conjugate gradient step.

tolfloat > 0 or None

Tolerance for the stopping criterion.

n_targets_batchint or None

Size of the batch for over targets during cross-validation. Used for memory reasons. If None, uses all n_targets at once.

random_stateint, or None

Random generator seed. Not used.

Returns

dual_weightsarray of shape (n_samples, n_targets)

Kernel ridge coefficients.

interceptarray of shape (n_targets,)

Intercept. Only returned when fit_intercept is True.