What exactly is the privacy problem in ML?
Within the scope of privacy in ML, there are two main concerns. The first is regarding the dataset itself – that is, how to collect it, how to keep it private, and how to prevent unauthorized access to sensitive information. The second is associated with the vulnerability of ML models to reveal the training data, which we will discuss in the next section. For now, let’s examine the issues related to dataset privacy in ML.
Copyright and intellectual property infringement
Copyright is a legal term that’s used to protect the ownership of intellectual property. It prevents or limits others from using your work without your permission. For example, if you take a photograph, record a video, or write a blog, your work is protected by copyright. Thus, others may not share, reproduce, or distribute your work without permission. Consequently, images, videos, text, or other information we see on the internet may have restrictive...
