CMPUT-664-Membership-Inference-Attacks-Against-Supervised-Learning-Models-on-Numeric-Data

Environment Setup

To setup all the necessary files and libraries which will be required, we have setup a requirements.txt file.

To install the essential libraries, use the following command:

pip install -r ./requirements.txt

Database Setup

For this project we have used 3 datasets which can be downloaded in ./Dataset folder. The three datasets which are being used are :

• Adult Income Dataset
• Heart Disease
• Pima Diabetes

Methodology

Step 1: Synthetic Data Generation

To generate the synthetic data, run the ./Code/Synthetic Data/Synthetic Dataset.ipynb using the following command

python -m ./Code/Synthetic Data/ Synthetic Dataset.ipynb

Step 2: Overfit Model Generation

The next step is to generate the overfitted model on the original dataset, the code used for that is available in ./Code/Overfitted Models/overfitted_model.ipynb, to run the python notebook use the following command:

python -m ./Code/Synthetic Data/overfitted_model.ipynb

Step 3 : Shadow Model Generation

Once the synthetic dataset and overfit model is complete, to generate the shadow models, run the ./Code/Shadow Models/.Shadow_Attack_{Dataset}, to run the python notebook use the following command :

python -m ./Code/Shadow Models/Shadow_Attack{Dataset}.ipynb

Evaluation

Census Data Set

Pima Diabetes Data Set

Heart Disease Data Set