Codebase for "Fair-GAIN" for fair ML predictions.
-
Updated
Mar 24, 2023 - Python
Codebase for "Fair-GAIN" for fair ML predictions.
Fairness-aware Data Mining
Code for the definition and testing of three new fairness-aware algorithms: Fair Decision Tree, Fair Genetic Pruning, and Fair LightGBM (FDT, FGP, FLGBM), completed for my Master's thesis.
Replicates the work from David Solans et al 2020 on poisoning algorithmic fairness.
Data Utility Improvement Experiment for DECAF
Demos for my blog, Ponder @ ponder.substack.com
"AdaFair: Cumulative Fairness Adaptive Boosting" algorithm (CIKM 2019) and its extension to other parity-based fairness notions (@Kais2022); Repository maintained by Vasileios Iosifidis.
Fair Classification with Gaussian Process (FCGP)
Repository of the paper "Towards a Human-Centred Fairness Analysis: From Binary to Multiclass and Multigroup Assessment in Graph Neural Network-Based Models for User Profiling Tasks"
Unofficial implementation of paper "Flexibly Fair Representation Learning by Disentanglement"
Fair decision making with unlabeled and labeled data
Evaluating ensemble performance in long-tailed datasets (Neurips 2023 Heavy Tails Workshop)
Robust Bayesian Recourse: a robust model-agnostic algorithmic recourse method (UAI'22)
Tensorflow implementation of Learning Not to Learn (CVPR 2019)
[TMLR 2024] Official implementation for "Understanding Fairness Surrogate Functions in Algorithmic Fairness"
Service to examine data processing pipelines (e.g., machine learning or deep learning pipelines) for uncertainty consistency (calibration), fairness, and other safety-relevant aspects.
Moment Decorrelation for constrained NN outputs with respect to protected attributes. (Decorrelated (flat), linear, quadratic, etc.)
Code related to the project: GetFair: Generalized Fairness Tuning of Classification Models
Add a description, image, and links to the fairness-ml topic page so that developers can more easily learn about it.
To associate your repository with the fairness-ml topic, visit your repo's landing page and select "manage topics."