From 88f14edb6a324072cfb22e3ae4f9c7bad531f511 Mon Sep 17 00:00:00 2001 From: Alexander Loosley Date: Sun, 10 Aug 2025 14:54:42 +0200 Subject: [PATCH 1/3] bibtex for citation --- README.md | 21 ++++++++++++++++++--- 1 file changed, 18 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index bab5bd5..2f4c04f 100644 --- a/README.md +++ b/README.md @@ -5,12 +5,27 @@ # FAIM -FAIM (FAir Interpolation Method), described in -[Beyond Incompatibility: Interpolation between Mutually -Exclusive Fairness Criteria in Classification Problems](https://arxiv.org/abs/2212.00469), +FAIM (FAir Interpolation Method) is a post-processing algorithm for achieving a combination of group-fairness criteria (equalized false positive rates, equalized false negative rates, group calibration). +## Citing +This work was [published in 2025](https://www.sciencedirect.com/science/article/pii/S0004370224002169) in the Journal of Artificial Intelligence. If you use FAIM or use ideas from this work to develop your own algorithms, please cite the paper: + +```bibtex +@article{ZEHLIKE2025104280, + title = {Beyond incompatibility: Trade-offs between mutually exclusive fairness criteria in machine learning and law}, + journal = {Artificial Intelligence}, + volume = {340}, + pages = {104280}, + year = {2025}, + issn = {0004-3702}, + doi = {https://doi.org/10.1016/j.artint.2024.104280}, + url = {https://www.sciencedirect.com/science/article/pii/S0004370224002169}, + author = {Meike Zehlike and Alex Loosley and Håkan Jonsson and Emil Wiedemann and Philipp Hacker} +} +``` + 🚧 **This README.md is under construction!** 🚧 ## Getting Started From 4f210a3e2b7a745d0f7d4bbdaa1a4151eb0e6ece Mon Sep 17 00:00:00 2001 From: Alexander Loosley Date: Sun, 10 Aug 2025 15:03:21 +0200 Subject: [PATCH 2/3] clear short introduction to faim --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 2f4c04f..ff7a77f 100644 --- a/README.md +++ b/README.md @@ -6,8 +6,8 @@ # FAIM FAIM (FAir Interpolation Method) -is a post-processing algorithm for achieving a combination of group-fairness criteria -(equalized false positive rates, equalized false negative rates, group calibration). +is a post-processing algorithm that can be used to balance between different group-fairness criteria. +At the moment, three group-fairness criteria have been implemented: equalized false positive rates, equalized false negative rates, and group calibration. See our [paper](#citing) for more details. ## Citing This work was [published in 2025](https://www.sciencedirect.com/science/article/pii/S0004370224002169) in the Journal of Artificial Intelligence. If you use FAIM or use ideas from this work to develop your own algorithms, please cite the paper: From 564e6c432d094da60d59995a8aa86a9a073a2c7b Mon Sep 17 00:00:00 2001 From: Alexander Loosley Date: Sun, 10 Aug 2025 15:08:43 +0200 Subject: [PATCH 3/3] update paper links from arxiv to AIJ --- README.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index ff7a77f..5c5aa9f 100644 --- a/README.md +++ b/README.md @@ -73,7 +73,7 @@ pip install faim #### Paper Implementation -The package and experiment CLI used to reproduce all the results in the [paper](https://arxiv.org/abs/2212.00469) +The package and experiment CLI used to reproduce all the results in the [paper](https://www.sciencedirect.com/science/article/pii/S0004370224002169) can be installed with: ```bash pip install "faim[experiment]" @@ -155,14 +155,14 @@ faim.predict(y_scores, sensitive_features) ### Paper Implementation -This section contains information for reproducing experiments in our [paper](https://arxiv.org/abs/2212.00469). +This section contains information for reproducing experiments in our [paper](https://www.sciencedirect.com/science/article/pii/S0004370224002169). Ensure the package has been installed with `[experiment]` extra requirements before continuing (see [Installation | Paper Implementation](#python-package))! Don't forget to restart your terminal before using the `faim` CLI in the steps below. #### Prepare Data -The CLI can be used to prepare any of the three datasets used in the [paper](https://arxiv.org/abs/2212.00469): +The CLI can be used to prepare any of the three datasets used in the [paper](https://www.sciencedirect.com/science/article/pii/S0004370224002169): ```bash faim-experiment --prepare-data DATASET ``` @@ -210,7 +210,7 @@ desired amount of the three fairness criteria that the system should achieve: Note, as discussed in the paper, thetas = 1,1,1 does not indicate that the system will simultaneously achieve all three (mutually incompatible) fairness criteria, but rather the result will be a compromise between all three. -See the [paper](https://arxiv.org/abs/2212.00469) for more details. +See the [paper](https://www.sciencedirect.com/science/article/pii/S0004370224002169) for more details. Finally, `PREPARED_DATA_FILEPATH` corresponds to the filepath of the prepared data.