Skip to content

Official implementation of "Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction"

License

Notifications You must be signed in to change notification settings

ArjunSubramonian/link_bias_amplification

Repository files navigation

Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction

This repository is the official implementation of Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction.

Requirements

To install requirements, use the provided requirements.txt:

pip install -r requirements.txt

Experiments

To run the experiments validating our theoretical analysis, simply run generate_plots.ipynb. To run the fairness experiments, run generate_fairness_plots.ipynb.

License

Some utilities code files are adapted from the PyTorch Geometric library, the FairGNN repository, and the Nifty repository; these files contain the appropriate copyright notices. All other code is written by the authors.

About

Official implementation of "Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published