C.3: Locally Private Graph Neural Networks

• Short Description: Federated training of Graph Neural Networks (GNNs) with Local Differential Privacy,  The code repository contains a python script, which can be used to train locally private GNNs, and a notebook to demonstrate the results.

• Reference, URL: https://www.ai4europe.eu/research/ai-catalog/locally-private-graph-neural-networks

• Applicable Business Category: Research (Cloud, Edge and Infrastructure)

• Application in relevant Projects/Initiatives: AI4Media

• Asset Type: Jupyter Notebook

• AI Breadth: N.A.

• Learning Ability: Graph Neural Networks

• Related technologies: Python script

• Applicable Research Area: Collaborative AI, Verifiable AI

• Applicable Technical Category: AI Ethics, Machine Learning

• License Information: MIT license (MIT)

• Related to circularity and sustainability: Yes

• Audience: Developers

Go back to the AI portfolio of reference implementations