"""File with Saliency algorithm explainer classes.
Based on https://github.com/pytorch/captum/blob/master/captum/attr/_core/saliency.py.
"""
from typing import Any
import torch
from captum._utils.typing import TargetType
from captum.attr import Saliency
from foxai.array_utils import validate_result
from foxai.explainer.base_explainer import Explainer
[docs]class SaliencyCVExplainer(Explainer):
"""Saliency algorithm explainer."""
[docs] def calculate_features(
self,
model: torch.nn.Module,
input_data: torch.Tensor,
pred_label_idx: TargetType = None,
abs_value: bool = True,
additional_forward_args: Any = None,
**kwargs, # pylint: disable = (unused-argument)
) -> torch.Tensor:
"""Generate model's attributes with Saliency algorithm explainer.
Args:
model: The forward function of the model or any
modification of it.
input_data: Input for which saliency
is computed. If forward_func takes a single tensor
as input, a single input tensor should be provided.
pred_label_idx: Output indices for
which gradients are computed (for classification cases,
this is usually the target class).
If the network returns a scalar value per example,
no target index is necessary.
For general 2D outputs, targets can be either:
- a single integer or a tensor containing a single
integer, which is applied to all input examples
- a list of integers or a 1D tensor, with length matching
the number of examples in inputs (dim 0). Each integer
is applied as the target for the corresponding example.
For outputs with > 2 dimensions, targets can be either:
- A single tuple, which contains #output_dims - 1
elements. This target index is applied to all examples.
- A list of tuples with length equal to the number of
examples in inputs (dim 0), and each tuple containing
#output_dims - 1 elements. Each tuple is applied as the
target for the corresponding example.
Default: None
abs_value: Returns absolute value of gradients if set
to True, otherwise returns the (signed) gradients if
False.
Default: True
additional_forward_args: If the forward function
requires additional arguments other than the inputs for
which attributions should not be computed, this argument
can be provided. It must be either a single additional
argument of a Tensor or arbitrary (non-tuple) type or a
tuple containing multiple additional arguments including
tensors or any arbitrary python types. These arguments
are provided to forward_func in order following the
arguments in inputs.
Note that attributions are not computed with respect
to these arguments.
Default: None
Returns:
The gradients with respect to each input feature.
Attributions will always be
the same size as the provided inputs, with each value
providing the attribution of the corresponding input index.
If a single tensor is provided as inputs, a single tensor is
returned. If a tuple is provided for inputs, a tuple of
corresponding sized tensors is returned.
Raises:
RuntimeError: if attribution has shape (0).
"""
saliency = Saliency(forward_func=model)
attributions = saliency.attribute(
input_data,
target=pred_label_idx,
abs=abs_value,
additional_forward_args=additional_forward_args,
)
validate_result(attributions=attributions)
return attributions