asaplib.kernel package

Submodules

asaplib.kernel.kernel_transforms module

Methods and functions to convert descriptors to kernels for samples

Kernels are measures of similarity, i.e. s(a, b) > s(a, c) if objects a and b are considered “more similar” than objects a and c. A kernel must also be positive semi-definite.

Essentially, for each pair of samples a and b we compute k(a,b) based on the coordinates of descriptors d(a) and d(b)

class asaplib.kernel.kernel_transforms.Descriptors_to_Kernels(k_spec_dict={})

Bases: object

add(k_spec, tag)

adding the specifications of a new kernel function :param k_spec: :type k_spec: a dictionary that specify which atomic descriptor to use

bind()

binds the objects that actually compute the kernels these objects need to have .transform() method to compute kernels from decriptor matrix [n_descriptors, n_samples]

compute(desc_a, desc_b=None)

compute the global descriptor vector for a frame from atomic contributions :param desc: design matrix :type desc: array-like, shape=[n_descriptors, n_samples]

Returns

k_mat – design matrix

Return type

array-like, shape=[n_samples, n_samples]

get_acronym()

pack()

class asaplib.kernel.kernel_transforms.Kernel_Function_Base(k_spec)

Bases: object

get_acronym()

transform(desc_a, desc_b)

class asaplib.kernel.kernel_transforms.Kernel_Function_Cosine(k_spec)

Bases: asaplib.kernel.kernel_transforms.Kernel_Function_Base

transform(desc_a, desc_b)

class asaplib.kernel.kernel_transforms.Kernel_Function_Linear(k_spec)

Bases: asaplib.kernel.kernel_transforms.Kernel_Function_Base

transform(desc_a, desc_b)

class asaplib.kernel.kernel_transforms.Kernel_Function_Polynomial(k_spec)

Bases: asaplib.kernel.kernel_transforms.Kernel_Function_Base

transform(desc_a, desc_b)

asaplib.kernel.ml_kernel_operations module

some operations for kernel and distance matrices

asaplib.kernel.ml_kernel_operations.distorho_quick(dis, delta)

asaplib.kernel.ml_kernel_operations.kerneltodis(kernel)

asaplib.kernel.ml_kernel_operations.kerneltodis_linear(kernel)

asaplib.kernel.ml_kernel_operations.kerneltorho(kernel, delta)

asaplib.kernel.ml_kernel_operations.normalizekernel(kernel)

Module contents