ICMKernelFactory

class baybe.surrogates.gaussian_process.components.kernel.ICMKernelFactory[source]

Bases: _MetaKernelFactory

A kernel factory that constructs an ICM kernel for transfer learning.

ICM: Intrinsic Coregionalization Model [BCW07]

Public methods

__init__([base_kernel_or_factory, ...])

Method generated by attrs for class ICMKernelFactory.

__call__(searchspace, objective, measurements)

Create a Gaussian process component for the given recommendation context.

Public attributes and properties

base_kernel_factory

The factory for the base kernel operating on numerical input features.

task_kernel_factory

The factory for the task kernel operating on the task indices.
__call__(searchspace: SearchSpace, objective: Objective, measurements: pd.DataFrame)[source]

Create a Gaussian process component for the given recommendation context.

Parameters:

  • searchspace (SearchSpace) – The optimization search space.

  • objective (Objective) – The optimization objective.

  • measurements (pd.DataFrame) – The available experimentation data.

Return type:

Kernel | GPyTorchKernel

Returns:

The constructed Gaussian process component.

__init__(base_kernel_or_factory: GPComponent | GPComponentFactoryProtocol = NOTHING, task_kernel_or_factory: GPComponent | GPComponentFactoryProtocol = NOTHING)

Method generated by attrs for class ICMKernelFactory.

For details on the parameters, see Public attributes and properties.

base_kernel_factory: GPComponentFactoryProtocol[Kernel]

The factory for the base kernel operating on numerical input features.

task_kernel_factory: GPComponentFactoryProtocol[Kernel]

The factory for the task kernel operating on the task indices.