Subsection: Risks in using PETs

[dsmits]

In the second paragraph of Federated Learning, I would emphasise that the gradient leakage is one example of a vulnerability. Maybe briefly also mention the others and that there might be still unknown attacks. I don't feel we should explain them in great detail as that would distract from the point of this chapter. We also could consider making this its own subsection.

• [ ] Reconstruction
• [ ] Differencing
• [ ] Deep Leakage from Gradients (DLG)
• [ ] Generative Adversarial Networks (GAN)
• [ ] Model Inversion
• [ ] Watermark Attacks

[dsmits]

Thinking to make it a callout box with a couple of links