About

This repository includes the experiments that accompany the paper "Assessing the Impact of Membership Inference Attacks on Classical Machine Learning Algorithms", submitted to 1st International Workshop on Emerging Technologies to Deploy Secure and Reliable Edge Computing Networks, Systems and Services (Go2Edge).

Dependencies

• Python
• Numpy, Pandas, Scikit-learn
• Adversarial Robustness Toolbox

Content

Datasets/ contains part of the datasets used.

Assessing-the-Impact-of-Membership-Inference-Attacks-on-Classical-Machine-Learning-Algorithms.ipynb contains the experiments performed.

Usage

Open the file with jupiter notebook or jupiter lab.

Datasets used:

1. Adult
2. Car Evaluation
3. Nursery
4. Cancer breast
5. MNIST
6. Titanic

Models evaluated:

1. Decision Tree (CART).
2. Bagging Classifier.
3. Random Forest.
4. Extra-Trees.
5. AdaBoost.
6. Gradient Boosting.
7. Logistic Regression.
8. Support Vector Classification.