privacytrustlab / ml_privacy_meter

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

version 0.1 release #97

Closed changhongyan123 closed 1 year ago

changhongyan123 commented 1 year ago

This PR is for the release of version 0.1 of our privacy meter software! ๐ŸŽ‰

This update includes the following improvements:

โœจ New features:

Example code for running the privacy meter with different games ๐ŸŽฎ Ways to save all models during each run ๐Ÿ“‚ Speed up the training of the models on PyTorch ๐Ÿš€ Seed up the efficiency of the privacy meter ๐Ÿ•ฐ๏ธ Time logging for each run โฑ๏ธ Configuration Yaml file for easy setup ๐Ÿ“„ White-box tutorial based on the gradient norm information ๐Ÿ“ Gradient and gradient norm signals added to the PyTorch model ๐Ÿ”ข Unit test added ๐Ÿงช End-to-end test for reproducing the results on the CIFAR10 dataset ๐Ÿ“Š Guidelines for potential contributors ๐Ÿค Slack channel set up for better communications ๐Ÿ’ฌ

๐Ÿ› Bug fixes:

Fixed bugs in previous tutorials ๐Ÿž