Note: this repository serves as a toolkit for the fairness project from the DTU special course: Responsible AI. For further information about this course, please contact us.
This repository provides some sample codes and tool functions for this project work. The purpose of providing this repository is to save time for building classifiers and dataloading, which means, your project could be built on a totally different framework and structure. There should be no limitation about the detailed implementation in this project.
Check here.
Please check this notebook: playground.ipynb
In this notebook, we will show examples of:
- Load the dataset and avoid the pitfall in joint-distribution imbalance.
- A simple resnet implementation with the dataset.
- Tricks to use some mitigation methods in
FairLearnwith multi-D input.
ninwe[at]dtu.dk