Machine Learning Phases and Privacy Threats/Attacks in Each Phase

In this chapter, we will provide a quick refresher on the different types of machine learning (ML): supervised, unsupervised, and reinforcement learning. We will also review the essential phases or pipelines of ML. You may already be familiar with these; if not, this chapter will serve as a foundational introduction.

Subsequently, we will delve into the crucial topic of privacy preservation within each phase of the ML process. Specifically, we will explore the importance of maintaining privacy in training data, input data, model storage, and inference/output data. Additionally, we will examine various privacy attacks that can occur in each phase, such as training data extraction attacks, model inversion attacks, and model inference attacks. Through detailed examples, we will gain an understanding of how these attacks function and discuss strategies to safeguard against them.

We will cover the following main topics...