quantus.functions.discretise_func module
This module holds a collection of explanation discretisation functions i.e., methods to split continuous explanation spaces into discrete counterparts.
- quantus.functions.discretise_func.floating_points(a: array, **kwargs) float
Rounds input to have n floating-points representation
- Parameters:
- a: np.ndarray
Numpy array with shape (x,).
- kwargs: optional
Keyword arguments.
n: integer Number of floating point digits.
- Returns:
- float
Returns the hash values of the resulting array.
- quantus.functions.discretise_func.rank(a: array, **kwargs) float
Calculates indices that would sort the array in order of importance.
- Parameters:
- a: np.ndarray
Numpy array with shape (x,).
- kwargs: optional
Keyword arguments.
- Returns:
- float
Returns the hash values of the resulting array.
- quantus.functions.discretise_func.sign(a: array, **kwargs) float
Calculates element-wise signs of the array.
- Parameters:
- a: np.ndarray
Numpy array with shape (x,).
- kwargs: optional
Keyword arguments.
- Returns:
- float
Returns the hash values of the resulting array.
- quantus.functions.discretise_func.top_n_sign(a: array, **kwargs) float
Calculates top n element-wise signs of the array.
- Parameters:
- a: np.ndarray
Numpy array with shape (x,).
- kwargs: optional
Keyword arguments.
n: integer Number of floating point digits.
- Returns:
- float
Returns the hash values of the resulting array.