apunkt/ai-privacy-toolkit
1
0
Fork 0
mirror of https://github.com/IBM/ai-privacy-toolkit.git synced 2026-06-05 14:55:13 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add dataset privacy risk assessment example notebook. (#73)

Browse source 
* Add dataset assessment notebook and reference to module from project README

Signed-off-by: Maya Anderson <mayaa@il.ibm.com>
This commit is contained in:
andersonm-ibm 2023-05-04 12:21:42 +03:00 • committed by abigailt
parent a4e07faf01
commit 98a7a078bb
4 changed files with 410 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
README.md
View file

@ -16,6 +16,9 @@ minimization principle in GDPR for ML models. It enables to reduce the amount of
personal data needed to perform predictions with a machine learning model, while still enabling the model personal data needed to perform predictions with a machine learning model, while still enabling the model
to make accurate predictions. This is done by by removing or generalizing some of the input features. to make accurate predictions. This is done by by removing or generalizing some of the input features.
The [**dataset assessment**](apt/risk/data_assessment/README.md) module implements a tool for privacy assessment of
synthetic datasets that are to be used in AI model training.
Official ai-privacy-toolkit documentation: https://ai-privacy-toolkit.readthedocs.io/en/latest/ Official ai-privacy-toolkit documentation: https://ai-privacy-toolkit.readthedocs.io/en/latest/
Installation: pip install ai-privacy-toolkit Installation: pip install ai-privacy-toolkit

1
apt/risk/data_assessment/dataset_attack.py
View file

@ -87,7 +87,6 @@ class DatasetAttackMembership(DatasetAttack):
labels = np.concatenate((np.zeros((len(non_member_probabilities),)), np.ones((len(member_probabilities),)))) labels = np.concatenate((np.zeros((len(non_member_probabilities),)), np.ones((len(member_probabilities),))))
results = np.concatenate((non_member_probabilities, member_probabilities)) results = np.concatenate((non_member_probabilities, member_probabilities))
svc_disp = RocCurveDisplay.from_predictions(labels, results) svc_disp = RocCurveDisplay.from_predictions(labels, results)
svc_disp.plot()
plt.plot([0, 1], [0, 1], color="navy", linewidth=2, linestyle="--", label='No skills') plt.plot([0, 1], [0, 1], color="navy", linewidth=2, linestyle="--", label='No skills')
plt.title('ROC curve') plt.title('ROC curve')
plt.savefig(f'{filename_prefix}{dataset_name}_roc_curve.png') plt.savefig(f'{filename_prefix}{dataset_name}_roc_curve.png')

402
notebooks/dataset_assessment_nursery.ipynb Normal file
View file

File diff suppressed because one or more lines are too long

5
requirements.txt
View file

@ -13,3 +13,8 @@ tensorflow==2.8.3
xgboost==1.7.2 xgboost==1.7.2
Pillow==9.3.0 Pillow==9.3.0
sortedcontainers==2.4.0 sortedcontainers==2.4.0
#notebooks
notebook
jupyter
ipywidgets