grakel.GraphHopper

class grakel.GraphHopper(n_jobs=None, normalize=False, verbose=False, kernel_type='linear')

Graph Hopper Histogram kernel as found in [FKP+13].

Parameters

kernel_typestr, tuple or function

For kernel_type of type:

• str can either be ‘linear’, ‘gaussian’, ‘bridge’.

• tuple can be of the form (‘gaussian’, mu) where mu is a number.

• function can be a function that takes two tuples of np.arrays for each graph corresponding to the M matrix and the attribute matrix and returns a number.

Attributes

metric_function

The base metric applied between features.

calculate_norm_bool

Defines if the norm of the attributes will be calculated (in order to avoid recalculation when using it with e.g. gaussian).

Methods

diagonal(self)	Calculate the kernel matrix diagonal of the fit/transformed data.
fit(self, X[, y])	Fit a dataset, for a transformer.
fit_transform(self, X)	Fit and transform, on the same dataset.
get_params(self[, deep])	Get parameters for this estimator.
initialize(self)	Initialize all transformer arguments, needing initialization.
pairwise_operation(self, x, y)	Graph Hopper kernel as proposed in [FKP+13].
parse_input(self, X)	Parse and check the given input for the Graph Hopper kernel.
set_params(self, \*\*params)	Call the parent method.
transform(self, X)	Calculate the kernel matrix, between given and fitted dataset.

Initialize an Graph Hopper kernel.

Attributes

X

Methods

diagonal(self)	Calculate the kernel matrix diagonal of the fit/transformed data.
fit(self, X[, y])	Fit a dataset, for a transformer.
fit_transform(self, X)	Fit and transform, on the same dataset.
get_params(self[, deep])	Get parameters for this estimator.
initialize(self)	Initialize all transformer arguments, needing initialization.
pairwise_operation(self, x, y)	Graph Hopper kernel as proposed in [FKP+13].
parse_input(self, X)	Parse and check the given input for the Graph Hopper kernel.
set_params(self, \*\*params)	Call the parent method.
transform(self, X)	Calculate the kernel matrix, between given and fitted dataset.

__init__(self, n_jobs=None, normalize=False, verbose=False, kernel_type='linear')

Initialize an Graph Hopper kernel.

diagonal(self)

Calculate the kernel matrix diagonal of the fit/transformed data.

Parameters

None.

Returns

X_diagnp.array

The diagonal of the kernel matrix between the fitted data. This consists of each element calculated with itself.

Y_diagnp.array

The diagonal of the kernel matrix, of the transform. This consists of each element calculated with itself.

fit(self, X, y=None)

Fit a dataset, for a transformer.

Parameters

Xiterable

Each element must be an iterable with at most three features and at least one. The first that is obligatory is a valid graph structure (adjacency matrix or edge_dictionary) while the second is node_labels and the third edge_labels (that fitting the given graph format). The train samples.

yNone

There is no need of a target in a transformer, yet the pipeline API requires this parameter.

Returns

selfobject

Returns self.

fit_transform(self, X)

Fit and transform, on the same dataset.

Parameters

Xiterable

Each element must be an iterable with at most three features and at least one. The first that is obligatory is a valid graph structure (adjacency matrix or edge_dictionary) while the second is node_labels and the third edge_labels (that fitting the given graph format). If None the kernel matrix is calculated upon fit data. The test samples.

yNone

There is no need of a target in a transformer, yet the pipeline API requires this parameter.

Returns

Knumpy array, shape = [n_targets, n_input_graphs]

corresponding to the kernel matrix, a calculation between all pairs of graphs between target an features

get_params(self, deep=True)

Get parameters for this estimator.

Parameters

deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns

paramsmapping of string to any

Parameter names mapped to their values.

initialize(self)

Initialize all transformer arguments, needing initialization.

pairwise_operation(self, x, y)

Graph Hopper kernel as proposed in [FKP+13].

Parameters

x, ytuple

Extracted features from parse_input.

Returns

kernelnumber

The kernel value.

parse_input(self, X)

Parse and check the given input for the Graph Hopper kernel.

Parameters

Xiterable

For the input to pass the test, we must have: Each element must be an iterable with at most three features and at least one. The first that is obligatory is a valid graph structure (adjacency matrix or edge_dictionary) while the second is node_labels and the third edge_labels (that fitting the given graph format).

Returns

outnp.array, shape=(len(X), n_labels)

A np array for frequency (cols) histograms for all Graphs (rows).

set_params(self, **params)

Call the parent method.

transform(self, X)

Calculate the kernel matrix, between given and fitted dataset.

Parameters

Xiterable

Each element must be an iterable with at most three features and at least one. The first that is obligatory is a valid graph structure (adjacency matrix or edge_dictionary) while the second is node_labels and the third edge_labels (that fitting the given graph format). If None the kernel matrix is calculated upon fit data. The test samples.

Returns

Knumpy array, shape = [n_targets, n_input_graphs]

corresponding to the kernel matrix, a calculation between all pairs of graphs between target an features

Bibliography

FKP+13(1,2,3,4)

Aasa Feragen, Niklas Kasenburg, Jens Petersen, Marleen de Bruijne, and Karsten Borgwardt. Scalable kernels for graphs with continuous attributes. In Advances in Neural Information Processing Systems, 216–224. 2013.