thupchnsky/sgc_unlearn
Certified (approximate) machine unlearning for simplified graph convolutional networks (SGCs) with theoretical guarantees (ICLR 2023)
This project provides methods to efficiently and reliably remove specific data points from graph-based machine learning models, specifically Simplified Graph Convolutional Networks (SGCs), while ensuring the model behaves as if the data was never learned. It takes an existing SGC model and a request to remove node features, an entire node, or an edge, then outputs an updated model. Data scientists, machine learning engineers, and researchers working with graph data would use this to manage data privacy or correct errors.
No commits in the last 6 months.
Use this if you need to remove specific data points (nodes, edges, or node features) from a Simplified Graph Convolutional Network model and require theoretical guarantees that the unlearning process is effective.
Not ideal if your machine learning model is not a Simplified Graph Convolutional Network or if you are working with unstructured data rather than graph-structured data.
Stars
20
Forks
1
Language
Python
License
MIT
Category
Last pushed
Feb 17, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/thupchnsky/sgc_unlearn"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Trusted-AI/adversarial-robustness-toolbox
Adversarial Robustness Toolbox (ART) - Python Library for Machine Learning Security - Evasion,...
bethgelab/foolbox
A Python toolbox to create adversarial examples that fool neural networks in PyTorch, TensorFlow, and JAX
cleverhans-lab/cleverhans
An adversarial example library for constructing attacks, building defenses, and benchmarking both
DSE-MSU/DeepRobust
A pytorch adversarial library for attack and defense methods on images and graphs
BorealisAI/advertorch
A Toolbox for Adversarial Robustness Research