mvpa2.kernels.np.ExponentialKernel

class mvpa2.kernels.np.ExponentialKernel(*args, **kwargs)

The Exponential kernel class.

Note that it can handle a length scale for each dimension for Automtic Relevance Determination.

Methods

compute_lml_gradient(alphaalphaT_Kinv, data)	Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel.
compute_lml_gradient_logscale(...)	Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel.
gradient(data1, data2)	Compute gradient of the kernel matrix.

Initialize instance of ExponentialKernel

Parameters:

length_scale : float or list(float), optional

The characteristic length-scale (or length-scales) of the phenomenon under investigation. Constraints: value must be convertible to type ‘float’, or value must be convertible to list(float). [Default: 1.0]

sigma_f : float, optional

Signal standard deviation. Constraints: value must be convertible to type ‘float’. [Default: 1.0]

enable_ca : None or list of str

Names of the conditional attributes which should be enabled in addition to the default ones

disable_ca : None or list of str

Names of the conditional attributes which should be disabled

Methods

compute_lml_gradient(alphaalphaT_Kinv, data)	Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel.
compute_lml_gradient_logscale(...)	Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel.
gradient(data1, data2)	Compute gradient of the kernel matrix.

compute_lml_gradient(alphaalphaT_Kinv, data)

Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel. Shorter formula. Allows vector of lengthscales (ARD) BUT THIS LAST OPTION SEEMS NOT TO WORK FOR (CURRENTLY) UNKNOWN REASONS.

compute_lml_gradient_logscale(alphaalphaT_Kinv, data)

Compute grandient of the kernel and return the portion of log marginal likelihood gradient due to the kernel. Shorter formula. Allows vector of lengthscales (ARD). BUT THIS LAST OPTION SEEMS NOT TO WORK FOR (CURRENTLY) UNKNOWN REASONS.

gradient(data1, data2)

Compute gradient of the kernel matrix. A must for fast model selection with high-dimensional data.