fix: fixed np.sum in art/defences/postprocessor/gaussian_noise.py

[GianlucaMega]

Signed-off-By: GianlucaMega megagianluca.mg@gmail.com

Description

The GaussianNoise postprocessing defence could not handle mini-batches due to a problem with the sums variable, in particular with the np.sum. The original instruction has been changed into np.sum(post_preds, axis=1, keepdims=True)

Fixes #1675

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

Please describe the tests that you ran to verify your changes. Consider listing any relevant details of your test configuration.

• tests/defences/test_gaussian_noise.py

Test Configuration:

• OS Kali Linux 2022.1
• Python version 3.9.7
• ART version or commit number 1.10.2
• TensorFlow / Keras / PyTorch / MXNet version TF:2.8.0, Keras:2.8.0, PyTorch:1.11.0+cu102, mx:1.8.0

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

[codecov-commenter]

Codecov Report

Merging #1684 (e50e827) into dev_1.10.2 (cb95baf) will increase coverage by 7.91%.
The diff coverage is 100.00%.

@@              Coverage Diff               @@
##           dev_1.10.2    #1684      +/-   ##
==============================================
+ Coverage       80.16%   88.08%   +7.91%     
==============================================
  Files             259      259              
  Lines           21349    21354       +5     
  Branches         3789     3791       +2     
==============================================
+ Hits            17114    18809    +1695     
+ Misses           3384     1599    -1785     
- Partials          851      946      +95     

Impacted Files	Coverage Δ
art/defences/postprocessor/gaussian_noise.py	96.42% <100.00%> (ø)
art/defences/preprocessor/mp3_compression.py	84.05% <0.00%> (-5.01%)	⬇️
art/defences/detector/poison/activation_defence.py	84.28% <0.00%> (-0.34%)	⬇️
art/estimators/classification/tensorflow.py	84.48% <0.00%> (+0.41%)	⬆️
art/estimators/classification/scikitlearn.py	90.86% <0.00%> (+0.66%)	⬆️
art/attacks/attack.py	92.00% <0.00%> (+0.79%)	⬆️
art/utils.py	75.45% <0.00%> (+0.83%)	⬆️
art/attacks/evasion/boundary.py	93.97% <0.00%> (+1.20%)	⬆️
art/estimators/classification/keras.py	88.48% <0.00%> (+4.93%)	⬆️
art/attacks/evasion/iterative_method.py	100.00% <0.00%> (+9.09%)	⬆️
... and 18 more

[beat-buesser]

Hi @GianlucaMega Thank you very much for you pull request, that's amazing! May I ask you for one more step, could you please sign-off the commit to pass the DCO check? Follow these steps:

1. Go to your local copy of your branch dev_1.10.2 where you created the commit
2. Run git rebase HEAD~1 --signoff
3. Force push your changes to overwrite your remote branch with git push --force-with-lease origin dev_1.10.2

[GianlucaMega]

Sorry for the inconvenience, I've fixed it!

[beat-buesser]

Hi @GianlucaMega Thank you very much for your help and congratulations to your first contribution to ART! Please let me know when you are interested in working on a new project for ART.

[GianlucaMega]

Thank you very much @beat-buesser for the support and guidance you gave me and congratulations for ART, amazing library!