https://github.com/csong27/membership-inference

Code for Membership Inference Attack against Machine Learning Models (in Oakland 2017)
https://github.com/csong27/membership-inference

machine-learning privacy

Last synced: 3 months ago
JSON representation

Code for Membership Inference Attack against Machine Learning Models (in Oakland 2017)

• Host: GitHub
• URL: https://github.com/csong27/membership-inference
• Owner: csong27
• Created: 2017-06-06T08:42:05.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-11-15T18:40:25.000Z (about 7 years ago)
• Last Synced: 2024-08-03T12:02:04.116Z (6 months ago)
• Topics: machine-learning, privacy
• Language: Python
• Homepage:
• Size: 7.81 KB
• Stars: 175
• Watchers: 3
• Forks: 63
• Open Issues: 10
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Membership Inference Attack against Machine Learning Models

This repository contains example of experiments for the paper Membership Inference Attack against Machine Learning Models (http://ieeexplore.ieee.org/document/7958568/). 

### Attack Experiment

python attack.py train_feat_file train_label_file

**train_feat_file** should be a text file where each line is a feature vector with floating point values separated by comma. 

**train_label_file** should be a text file where each line is a label with integer value. 

Once data is loaded, we will split the data for training target model and shadow models and save the split to disk. Then we will train the target model as well as shadow models. Finally attack model can be trained with predictions from shadow models and test on the target model.