Class NoisePerturbationSensitivity

source code

This is a FeaturewiseDatasetMeasure that uses a scalar DatasetMeasure and selective noise perturbation to compute a sensitivity map.

First the scalar DatasetMeasure computed using the original dataset. Next the data measure is computed multiple times each with a single feature in the dataset perturbed by noise. The resulting difference in the scalar DatasetMeasure is used as the sensitivity for the respective perturbed feature. Large differences are treated as an indicator of a feature having great impact on the scalar DatasetMeasure.

The computed sensitivity map might have positive and negative values!

Nested Classes

[hide private]

Inherited from misc.state.ClassWithCollections: __metaclass__

Instance Methods

[hide private]

__init__(self, datameasure, noise=N.random.normal)
Cheap initialization.

source code

_call(self, dataset)
Compute the sensitivity map.

source code

Inherited from base.FeaturewiseDatasetMeasure: __repr__, combiner

Inherited from base.FeaturewiseDatasetMeasure (private): _postcall

Inherited from base.DatasetMeasure: __call__, null_dist, transformer, untrain

Inherited from misc.state.ClassWithCollections: __getattribute__, __new__, __setattr__, __str__, reset

Inherited from object: __delattr__, __format__, __hash__, __reduce__, __reduce_ex__, __sizeof__, __subclasshook__

Class Variables

[hide private]

Inherited from base.FeaturewiseDatasetMeasure: base_sensitivities

Inherited from base.DatasetMeasure: __doc__, null_prob, null_t, raw_results

Inherited from misc.state.ClassWithCollections: _DEV__doc__, descr

Properties

[hide private]

Inherited from object: __class__

Method Details

[hide private]

init(self, datameasure, noise=N.random.normal)
(Constructor)

source code

Cheap initialization.

Parameters:

combiner - The combiner is only applied if the computed featurewise dataset measure is more than one-dimensional. This is different from a transformer, which is always applied. By default, the sum of absolute values along the second axis is computed.

Overrides: object.__init__

Parameters:

datameasure: Datameasure that is used to quantify the effect of: noise perturbation.
noise: Functor to generate noise. The noise generator has to return: an 1d array of n values when called the size=n keyword argument. This is the default interface of the random number generators in NumPy's random module.

_call(self, dataset)

source code

Compute the sensitivity map.

Returns a 1d array of sensitivities for all features in dataset.

Overrides: base.DatasetMeasure._call

Home

Trees

Indices

Help

Class NoisePerturbationSensitivity

__init__(self, datameasure, noise=N.random.normal) (Constructor)

_call(self, dataset)

init(self, datameasure, noise=N.random.normal)
(Constructor)