Support regression models in several inference attacks

[GalHorowitz]

Description

• New Scikitlearn regression wrapper, ScikitlearnDecistionTreeRegressor moved to art.estimators.regression
• Added a fixture for the diabetes regression dataset from scitkitlearn
• _estimator_requirements now supports tuples to represent unions
• MembershipInferenceBlackBox now supports regression models (when input_type='loss')
• AttributeInferenceBlackBox and MembershipInferenceBlackBox now support regression models
• Two new notebooks to demonstrate membership and attribute inference on regression models
• New attribute inference baseline that uses the true label, used for evaluating attribute inference attacks

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

All unit tests pass.

Test Configuration:

• OS: MacOS 11.4
• Python version: 3.8.11

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 #1272 (7863d30) into dev_1.8.0 (bc6aa61) will decrease coverage by 15.47%.
The diff coverage is 69.67%.

@@              Coverage Diff               @@
##           dev_1.8.0    #1272       +/-   ##
==============================================
- Coverage      81.30%   65.83%   -15.48%     
==============================================
  Files            227      229        +2     
  Lines          19300    19488      +188     
  Branches        3502     3539       +37     
==============================================
- Hits           15691    12829     -2862     
- Misses          2309     5580     +3271     
+ Partials        1300     1079      -221     

Impacted Files	Coverage Δ
...ference/attribute_inference/true_label_baseline.py	26.66% <26.66%> (ø)
...acks/inference/attribute_inference/meminf_based.py	76.62% <50.00%> (-10.22%)	⬇️
art/estimators/regression/scikitlearn.py	73.40% <73.40%> (ø)
...attacks/inference/attribute_inference/black_box.py	71.42% <75.00%> (-16.64%)	⬇️
...ttacks/inference/membership_inference/black_box.py	78.42% <82.75%> (-9.15%)	⬇️
art/estimators/scikitlearn.py	91.66% <90.47%> (-8.34%)	⬇️
art/utils.py	73.40% <94.73%> (-0.69%)	⬇️
art/attacks/attack.py	91.85% <100.00%> (-1.85%)	⬇️
.../attacks/inference/attribute_inference/__init__.py	100.00% <100.00%> (ø)
...ute_inference/white_box_lifestyle_decision_tree.py	77.77% <100.00%> (+0.41%)	⬆️
... and 49 more

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging 4129a2f into e637029 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging 81cb991 into e637029 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[abigailgold]

@beat-buesser Are you sure that ScikitlearnDecisionTreeRegressor should be used within the get_trees method of ScikitlearnGradientBoostingClassifier?

[beat-buesser]

@abigailgold Yes, I think that should be correct. It looks like the trees of the ensemble in sklearn.ensemble.GradientBoostingClassifier are of type DecisionTreeRegressor in https://github.com/scikit-learn/scikit-learn/blob/2beed55847ee70d363bdbfe14ee4401438fba057/sklearn/ensemble/_gb.py#L195

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging 868f965 into e637029 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging edbc1af into e637029 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging 434dbf6 into e637029 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[lgtm-com]

This pull request introduces 5 alerts and fixes 1 when merging 961d1ef into a3525e9 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called

[lgtm-com]

This pull request introduces 5 alerts and fixes 2 when merging 7b84061 into a3525e9 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called
• 1 for Missing call to `__init__` during object initialization

[lgtm-com]

This pull request introduces 5 alerts and fixes 2 when merging 7863d30 into bc6aa61 - view on LGTM.com

new alerts:

• 4 for Module is imported with 'import' and 'import from'
• 1 for Module is imported more than once

fixed alerts:

• 1 for Non-callable called
• 1 for Missing call to `__init__` during object initialization

[beat-buesser]

Hi @GalHorowitz Thank you very much for this great pull request adding support for regression models in inference attacks!