"""File with Guided Backpropagation algorithm explainer class.
Based on https://github.com/pytorch/captum/blob/master/captum/attr/_core/guided_backprop_deconvnet.py.
"""
from abc import abstractmethod
from typing import Any
import torch
from captum._utils.typing import TargetType
from captum.attr import GuidedBackprop
from foxai.array_utils import validate_result
from foxai.explainer.base_explainer import Explainer
from foxai.explainer.computer_vision.model_utils import modify_modules
[docs]class BaseGuidedBackpropCVExplainer(Explainer):
"""Base Guided Backpropagation algorithm explainer."""
[docs] @abstractmethod
def create_explainer(
self,
model: torch.nn.Module,
**kwargs,
) -> GuidedBackprop:
"""Create explainer object.
Returns:
Explainer object.
"""
[docs] def calculate_features(
self,
model: torch.nn.Module,
input_data: torch.Tensor,
pred_label_idx: TargetType = None,
additional_forward_args: Any = None,
**kwargs,
) -> torch.Tensor:
"""Generate model's attributes with Guided Backpropagation algorithm explainer.
Args:
model: The forward function of the model or any
modification of it.
input_data: Input for which
attributions are computed. If forward_func takes a single
tensor as input, a single input tensor should be provided.
If forward_func takes multiple tensors as input, a tuple
of the input tensors should be provided. It is assumed
that for all given input tensors, dimension 0 corresponds
to the number of examples (aka batch size), and if
multiple input tensors are provided, the examples must
be aligned appropriately.
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
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 guided backprop 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).
"""
guided_backprop = self.create_explainer(model=model)
attributions = guided_backprop.attribute(
input_data,
target=pred_label_idx,
additional_forward_args=additional_forward_args,
)
validate_result(attributions=attributions)
return attributions
[docs]class GuidedBackpropCVExplainer(BaseGuidedBackpropCVExplainer):
"""Guided Backpropagation algorithm explainer."""
[docs] def create_explainer(
self,
model: torch.nn.Module,
**kwargs,
) -> GuidedBackprop:
"""Create explainer object.
Returns:
Explainer object.
"""
return GuidedBackprop(model=modify_modules(model))