Fisher Information Loss

This repository contains code that can be used to reproduce the experimental results presented in the paper:

Awni Hannun, Chuan Guo and Laurens van der Maaten. Measuring Data Leakage in Machine-Learning Models with Fisher Information. arXiv 2102.11673, 2021.

Installation

The code requires Python 3.7+, PyTorch 1.7.1+, and torchvision 0.8.2+.

Create an Anaconda environment and install the dependencies:

conda create --name fil
conda activate fil
conda install -c pytorch pytorch torchvision
pip install gitpython numpy

Usage

The script fisher_information.py computes the per-example FIL for the given dataset and model. An example run is:

python fisher_information.py \
    --dataset mnist \
    --model least_squares

To see usage options for the script run:

python fisher_information.py --help

Other scripts in the repository are:

• reweighted.py : Run the iteratively reweighted Fisher information loss (IRFIL) algorithm.
• model_inversion.py : Attribute inversion experiments for a non-private model.
• private_model_inversion.py : Attribute inversion experiments for a private model.
• test_jacobians.py : Unit tests.

To run the full set of experiments in the accompanying paper:

cd scripts/ && ./run_experiments.sh

Citing this Repository

If you use the code in this repository, please cite the following paper:

@inproceedings{hannun2021fil,
  title={Measuring Data Leakage in Machine-Learning Models with Fisher
    Information},
  author={Hannun, Awni and Guo, Chuan and van der Maaten, Laurens},
  booktitle={Conference on Uncertainty in Artificial Intelligence},
  year={2021}
}

License

This code is released under a CC-BY-NC 4.0 license. Please see the LICENSE file for more information.

Please review Facebook Open Source Terms of Use and Privacy Policy.