Set _device parameter automatically in PreprocessorPyTorch (fixes #1442)

[nilakshdas]

Description

The _device parameter is not automatically initialized in a PreprocessorPyTorch object. It is referenced in the PreprocessorPyTorch.device property, which in turn is used in the PreprocessorPyTorch.__call__ method. This PR alleviates the requirement for child classes to set the _device parameter and reduces code redundancy.

Fixes #1442.

Type of change

Please check all relevant options.

• Improvement (non-breaking)
• Bug fix (non-breaking)
• New feature (non-breaking)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Testing

$ pytest --framework=pytorch \
    tests/defences/preprocessor \
    tests/preprocessing/audio \
    tests/preprocessing/test_standardisation_mean_std.py
================================================================= test session starts =================================================================
platform linux -- Python 3.6.9, pytest-5.4.1, py-1.11.0, pluggy-0.13.1
rootdir: /workspaces/adversarial-robustness-toolbox, inifile: setup.cfg
plugins: mock-3.2.0, pep8-1.0.6, typeguard-2.13.3, anyio-3.4.0, cov-2.10.1
collected 133 items                                                                                                                                   

tests/defences/preprocessor/test_inverse_gan.py s                                                                                               [  0%]
tests/defences/preprocessor/test_jpeg_compression.py ssssssssssssssssssssss                                                                     [ 17%]
tests/defences/preprocessor/test_mp3_compression.py ..sssssssss                                                                                 [ 25%]
tests/defences/preprocessor/test_mp3_compression_pytorch.py .......                                                                             [ 30%]
tests/defences/preprocessor/test_preprocessor.py ........                                                                                       [ 36%]
tests/defences/preprocessor/test_resample.py sssss                                                                                              [ 40%]
tests/defences/preprocessor/test_spatial_smoothing.py sssssssssssss                                                                             [ 50%]
tests/defences/preprocessor/test_spatial_smoothing_pytorch.py x......xx......                                                                   [ 61%]
tests/defences/preprocessor/test_spatial_smoothing_tensorflow.py sssssssssssssss                                                                [ 72%]
tests/defences/preprocessor/test_video_compression.py sss...                                                                                    [ 77%]
tests/defences/preprocessor/test_video_compression_pytorch.py .......                                                                           [ 82%]
tests/preprocessing/audio/test_l_filter.py sssssssss                                                                                            [ 89%]
tests/preprocessing/audio/test_l_filter_pytorch.py ssssss.                                                                                      [ 94%]
tests/preprocessing/test_standardisation_mean_std.py sss..ss                                                                                    [100%]

====================================================== 42 passed, 88 skipped, 3 xfailed in 7.77s ======================================================

Test Configuration:

• OS: Ubuntu 18.04.4 LTS
• Python version: Python 3.6.9
• ART version or commit number: 4bf284b
• PyTorch: 1.5.0

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[lgtm-com]

This pull request introduces 1 alert when merging 5c84b3a into 09e0e19 - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

[codecov-commenter]

Codecov Report

Merging #1444 (1377522) into dev_1.9.0 (a7c1f7b) will decrease coverage by 0.03%.
The diff coverage is 100.00%.

@@              Coverage Diff              @@
##           dev_1.9.0    #1444      +/-   ##
=============================================
- Coverage      90.31%   90.27%   -0.04%     
=============================================
  Files            240      240              
  Lines          19682    19672      -10     
  Branches        3490     3490              
=============================================
- Hits           17775    17759      -16     
- Misses          1110     1114       +4     
- Partials         797      799       +2     

Impacted Files	Coverage Δ
...t/defences/preprocessor/mp3_compression_pytorch.py	94.59% <100.00%> (-0.41%)	⬇️
art/defences/preprocessor/preprocessor.py	78.26% <100.00%> (+1.89%)	⬆️
...defences/preprocessor/spatial_smoothing_pytorch.py	87.65% <100.00%> (-0.45%)	⬇️
...defences/preprocessor/video_compression_pytorch.py	94.73% <100.00%> (-0.39%)	⬇️
art/preprocessing/audio/l_filter/pytorch.py	100.00% <100.00%> (ø)
.../preprocessing/standardisation_mean_std/pytorch.py	100.00% <100.00%> (ø)
art/attacks/evasion/brendel_bethge.py	86.27% <0.00%> (-1.64%)	⬇️
art/attacks/evasion/boundary.py	92.72% <0.00%> (-1.22%)	⬇️

[lgtm-com]

This pull request introduces 1 alert when merging 5c84b3a into 24d7062 - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

[beat-buesser]

Hi @nilakshdas Thank you very much! This is a great pull request! It looks like pylint found that we can now remove some of the imports of torch but afterwards the pull request will be ready to be merged for ART 1.9.

[nilakshdas]

Hi @beat-buesser, I've fixed the import issues identified by pylint. Thanks so much for your guidance in my first PR to ART!

[lgtm-com]

This pull request introduces 1 alert and fixes 4 when merging 1e0f824 into df2e613 - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

fixed alerts:

• 4 for Module is imported more than once

[lgtm-com]

This pull request introduces 1 alert and fixes 5 when merging a7b5c93 into df2e613 - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

fixed alerts:

• 5 for Module is imported more than once

[lgtm-com]

This pull request introduces 1 alert and fixes 5 when merging 1377522 into a7c1f7b - view on LGTM.com

new alerts:

• 1 for Module is imported more than once

fixed alerts:

• 5 for Module is imported more than once