Deep Learning with Differential Privacy

Simple implementation of Deep Learning (DL) with Differential Privacy (DP).

Requirements

• torch 1.12.1
• functorch 0.2.1
• numpy 1.16.2
• opacus 1.3.0

Usage

1. Execute run_model.py -cf dp_sgd_config.json

Model parameters

dp_sgd_config.json

{
    "data_name": "mnist",
    "epochs": 100,
    "batch_size": 128, 
    "lr": 0.001, 
    "epsilon": 6, 
    "delta": 1e-05,
    "clipping_norm": 16.0, 
    "q": 0.05
}

Reference

[1] Abadi, Martin, et al. Deep learning with differential privacy. Proceedings of the 2016 ACM SIGSAC conference on computer and communications security. 2016.