privacytrustlab / ml_privacy_meter

Privacy Meter: An open-source library to audit data privacy in statistical and machine learning algorithms.
MIT License
557 stars 99 forks source link

Add reference membership inference attack #59

Closed amad-person closed 2 years ago

amad-person commented 2 years ago

With a tutorial script to demonstrate usage.

Thoughts on future improvements for a modular API:

  1. Dataset class: logic for data handling should be encapsulated
  2. Model class: so users can customize query interface for their models
  3. Attack class: some logic can be defined in a parent class, other attacks can extend this class + users can define their own attacks