[FEEDBACK]: Description of ML04 Membership Inference Attack

[dferri]

Type

General Feedback

What would you like to report?

www-project-machine-learning-security-top-10/docs/ML04_2023-Membership_Inference_Attack.md

Line 27 in f1cf662

Membership inference attacks occur when an attacker manipulates the model's

Here, the documentation states that an attacker manipulates the model’s training data, but from my understanding the objective of a membership inference attack is to "[...] predict whether or not a particular example was contained in the model’s training dataset.", so the attacker shouldn't have access to the training data.

I can create a pull request to update the documentation. Let me know if you'd like me to proceed.

Code of Conduct

• I agree to follow this project's Code of Conduct