niteshade.defence.Distance_metric

class niteshade.defence.Distance_metric(type='Eucleidian')

Bases: object

A Distance_metric class for the feasibleset defender. Allows to define custom distance metrics for feasibleset defender distance calculation. For user implemented custom Distance_metric objects, need to have a .distance method where a float is returned.

Parameters: type (string) – The type of the distance metric. This will be returned for informative purposes when .distance_metric is called for feasibleset defender.

__init__(type='Eucleidian') → None: Constructor method of FeasibleSetDefender class. Default Distance_metric is Eucleidian distance.

Methods

`__init__`([type])	Constructor method of FeasibleSetDefender class.
`distance`(input_1, Input_2)	Calculates the distance between 2 input points

distance(input_1, Input_2)

Calculates the distance between 2 input points: Currently only Eucleidian (l2 norm) distance metric is implemented off-the-shelf

Parameters

input_1 (np.ndarray) – point_1 data (shape data dimensionality).
input_2 (np.ndarray) – point_2 data (shape data dimensionality).

Returns

distance between the 2 input points.

Return type

distance (float)