Tracking PII ex-filtration: Exploring decision tree and neural network with explainable AI

The swift evolution of digital technology has led to the creation of devices that have substantially simplified people’s lives. Consequently, individuals are spending more and more time engaging with the internet. The impact of this has been positive as well as negative for users. Users’ interaction with smart-phone applications generates a large volume of sensitive personal data, which is stored in the device itself. The problem arises when data is ex-filtrated from the device without the user’s knowledge. Consequently, there is a dire need to detect any leak of user information from devices. In this view, we propose variations in machine learning frameworks that improve the state-of-the-art framework’s ability to detect users’ personal information disclosure with greater accuracy and precision. In addition, our framework integrates the XAI algorithm SHAP to enhance the interpretability and comprehension of the black-box approach when detecting sensitive data reliably. Our experimental findings validate that XAI offers improved explanations and responsiveness, making it a more user-friendly and transparent solution.