Customized Client and Dataset

I am very new to Flower framework. I want to know if we can implement k-anonymity on a few clients and local differential privacy on the others with different values of parameters using Flower.
If yes, if someone could please explain how can I do that.

1 Like

Hey, thanks for the question.

Yes, it is possible. First, you need to define your clients and partition their data. For clients where you want to implement k-anonymity, apply it to their data before using it for training. Additionally, for clients where you want to use Local DP, utilize the existing mod.

These links may be helpful:

Let me know if you have more questions.

1 Like