EnsembleXAI.Normalization module

EnsembleXAI.Normalization.mean_var_normalize(explanation_tensor, eps=1e-25)

Normalize explanations using mean and variance.

Parameters:

explanation_tensor (Tensor) – Explanations in the form of a tensor.
eps (float, optional) – Small constant to avoid division by zero, by default 1e-25.

Returns:

Normalized explanations.

Return type:

Tensor

See also

median_iqr_normalize: Normalize explanations using median and interquartile range.
second_moment_normalize: Normalize explanations using the second moment.

Examples

>>> import torch
>>> explanation = torch.randn(1, 3, 64, 64)  # Input explanation tensor
>>> normalized_explanation = mean_var_normalize(explanation)

EnsembleXAI.Normalization.median_iqr_normalize(explanation_tensor, eps=1e-25)

Normalize explanations using median and interquartile range.

Parameters:

explanation_tensor (Tensor) – Explanations in the form of a tensor.
eps (float, optional) – Small constant to avoid division by zero, by default 1e-25.

Returns:

Normalized explanations.

Return type:

Tensor

See also

mean_var_normalize: Normalize explanations using mean and variance.
second_moment_normalize: Normalize explanations using the second moment.

Examples

>>> import torch
>>> explanation = torch.randn(1, 3, 64, 64)  # Input explanation tensor
>>> normalized_explanation = median_iqr_normalize(explanation)

EnsembleXAI.Normalization.second_moment_normalize(explanation_tensor, eps=1e-25)

Normalize explanations using the second moment.

Parameters:

explanation_tensor (Tensor) – Explanations in the form of a tensor.
eps (float, optional) – Small constant to avoid division by zero, by default 1e-25.

Returns:

Normalized explanations.

Return type:

Tensor