• Stars
    star
    138
  • Rank 264,508 (Top 6 %)
  • Language
    Python
  • License
    MIT License
  • Created over 4 years ago
  • Updated about 1 year ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Machine Unlearning with SISA

Lucas Bourtoule, Varun Chandrasekaran, Christopher Choquette-Choo, Hengrui Jia, Adelin Travers, Baiwu Zhang, David Lie, Nicolas Papernot

This repository contains the core code used in the SISA experiments of our Machine Unlearning paper along with some example scripts.

You can start runing experiments by having a look at the readme in the purchase example folder at example-scripts/purchase-sharding.

sisa.py is the script that trains a given shard. It should be run as many times as the number of shards.

Citing this work

If you use this repository for academic research, you are highly encouraged (though not required) to cite our paper:

@inproceedings{bourtoule2021machine,
  title={Machine Unlearning},
  author={Lucas Bourtoule and Varun Chandrasekaran and Christopher Choquette-Choo and Hengrui Jia and Adelin Travers and Baiwu Zhang and David Lie and Nicolas Papernot},
  booktitle={Proceedings of the 42nd IEEE Symposium on Security and Privacy},
  year={2021}
}

More Repositories

1

cleverhans

An adversarial example library for constructing attacks, building defenses, and benchmarking both
Jupyter Notebook
6,168
star
2

entangled-watermark

Python
44
star
3

Proof-of-Learning

Python
43
star
4

dataset-inference

[ICLR'21] Dataset Inference for Ownership Resolution in Machine Learning
Jupyter Notebook
28
star
5

capc-iclr

CaPC is a method that enables collaborating parties to improve their own local heterogeneous machine learning models in a setting where both confidentiality and privacy need to be preserved to prevent explicit and implicit sharing of private data.
Python
26
star
6

unrolling-sgd

code release for "Unrolling SGD: Understanding Factors Influencing Machine Unlearning" published at EuroS&P'22
Python
22
star
7

verifiable-unlearning

Python
15
star
8

model-extraction-iclr

Python
12
star
9

DeCaPH

Python
7
star
10

deepfake_attribution

Python
5
star
11

Zest-Model-Distance

Python
5
star
12

DatasetInferenceForSelfSupervisedModels

Python
5
star
13

capc-demo

capc-demo
Jupyter Notebook
4
star
14

FRAUD-Detect

Official implementation of Washing The Unwashable : On The (Im)possibility of Fairwashing Detection, NeurIPS 2022
Python
4
star
15

ssl-attacks-defenses

On the Difficulty of Defending Self-Supervised Learning against Model Extraction
Python
3
star
16

Gradients-Look-Alike-Sensitivity-is-Often-Overestimated-in-DP-SGD

Python
2
star
17

Forging

Jupyter Notebook
1
star
18

monte-carlo-adv

Python
1
star
19

private-multi-winner-voting

New privacy-preserving mechanisms for multi-label classification.
Python
1
star
20

FairFeedbackLoops

Python
1
star
21

PrivatePrompts

Code for the differential learning algorithms for soft and discrete prompts.
Python
1
star