graphdot.kernel.fix module

class graphdot.kernel.fix.Exponentiation(kernel, xi=1.0, xi_bounds=(0.1, 20.0))

Bases: object

Raises a kernel to some exponentiation. \(k_\mathrm{exponentiation}(x, y) = k(x, y)^\xi\).

Parameters:

• kernel (object) – The graph kernel to be exponentiated.
• xi (float) – The exponent to be raises.
• xi_bounds ((float, float)) – The range of the exponents to be searched during hyperparameter optimization.

__call__(X, Y=None, eval_gradient=False, **options)

Normalized outcome of :py:`self.kernel(X, Y, eval_gradient, **options)`.

Parameters:that of the graph kernel object. (Inherits) – Returns: Return type:Inherits that of the graph kernel object.

bounds

clone_with_theta(theta)

diag(X, **options)

Normalized outcome of :py:`self.kernel.diag(X, **options)`.

Parameters:that of the graph kernel object. (Inherits) – Returns: Return type:Inherits that of the graph kernel object.

hyperparameter_bounds

hyperparameters

theta

class graphdot.kernel.fix.Normalization(kernel)

Bases: object

Normalizes a kernel using the cosine of angle formula: \(k_\mathrm{normalized}(x, y) = \frac{k(x, y)}{\sqrt{k(x, x)k(y, y)}}\).

Parameters:kernel (object) – The kernel to be normalized.

__call__(X, Y=None, eval_gradient=False, **options)

Normalized outcome of :py:`self.kernel(X, Y, eval_gradient, **options)`.

Parameters:that of the graph kernel object. (Inherits) – Returns: Return type:Inherits that of the graph kernel object.

bounds

clone_with_theta(theta)

diag(X, eval_gradient=False, **options)

Normalized outcome of :py:`self.kernel.diag(X, **options)`.

Parameters:that of the graph kernel object. (Inherits) – Returns: Return type:Inherits that of the graph kernel object.

hyperparameter_bounds

hyperparameters

theta