Code for FAccT paper: Fairness-aware Model-agnostic Positive and Unlabeled Learning Paper.
- numpy 1.19.2
- scikit-learn 0.23.2
- pulearn 0.0.7
- src/gen_syn.py: generate the synthetic dataset
- src/dataloader.py: load real data sets
- src/PUL.py: train and test our proposed framework
We have indicated how each step in All.1 in our paper corresponds to the code in the comments. To reproduce the experimental results in Table 1, simply run python PUL.py . The datasets can be downloaded from the weblink in the paper. All the carefully tuned parameters have been specified in the paper.