Differential Privacy

Man-in-the-Middle Attack

We use two datasets for this assessment: datasets/synthetic_NHANES.csv as the synthetic data, and datasets/NHANES.csv as the original dataset.

In this code, we apply the PETWorks API PETValidation(synthetic, original, "MIATest"), using the synthetic and original datasets as parameters. This function is designed to determine whether the data is processed without differential privacy or might potentially use differential privacy.

After the assessment, we use the report(result, "json") function to process and display the results. This function takes the evaluation result and the format parameter “json”. It outputs the evaluation in JSON format, with the result indicating whether the dataset is processed without differential privacy or might potentially use differential privacy.

Example: validateDP.py

from PETWorks import PETValidation, report

synthetic = "datasets/synthetic_NHANES.csv"
original = "datasets/NHANES.csv"

result = PETValidation(synthetic, original, "MIATest")
report(result, "json")

Execution Result

$ python3 validateDP.py
{
    "Does the data processed with differential privacy": "Possibly Yes"
}