niteshade.defence.OutlierDefender

class niteshade.defence.OutlierDefender(initial_dataset_x, initial_dataset_y)

Bases: niteshade.defence.Defender

Abstractclass for defenders that use a outlier filtering strategy.

Parameters

• initial_dataset_x (np.ndarray, torch.Tensor) – point data (shape (batch_size, data dimensionality)).

• initial_dataset_y (np.ndarray, torch.Tensor) – label data (shape (batch_size, )).

__init__(initial_dataset_x, initial_dataset_y) → None

Initialise the OutlierDefender class using a initial dataset.

Methods

__init__(initial_dataset_x, initial_dataset_y)	Initialise the OutlierDefender class using a initial dataset.
defend()	Checks if the .defend method is implemented.