The AI Fairness 360 toolkit is an open-source library to help detect and remove bias in machine learning models. The AI Fairness 360 Python package includes a comprehensive set of metrics for datasets and models to test for biases, explanations for these metrics, and algorithms to mitigate bias in datasets and models.
The AI Fairness 360 interactive experience provides a gentle introduction to the concepts and capabilities. The tutorials and other notebooks offer a deeper, data scientist-oriented introduction. The complete API is also available.
Being a comprehensive set of capabilities, it may be confusing to figure out which metrics and algorithms are most appropriate for a given use case. To help, we have created some guidance material that can be consulted.
We have developed the package with extensibility in mind. This library is still in development. We encourage the contribution of your metrics, explainers, and debiasing algorithms.
Get in touch with us on Slack (invitation here)!
- Optimized Preprocessing (Calmon et al., 2017)
- Disparate Impact Remover (Feldman et al., 2015)
- Equalized Odds Postprocessing (Hardt et al., 2016)
- Reweighing (Kamiran and Calders, 2012)
- Reject Option Classification (Kamiran et al., 2012)
- Prejudice Remover Regularizer (Kamishima et al., 2012)
- Calibrated Equalized Odds Postprocessing (Pleiss et al., 2017)
- Learning Fair Representations (Zemel et al., 2013)
- Adversarial Debiasing (Zhang et al., 2018)
- Meta-Algorithm for Fair Classification (Celis et al.. 2018)
- Comprehensive set of group fairness metrics derived from selection rates and error rates
- Comprehensive set of sample distortion metrics
- Generalized Entropy Index (Speicher et al., 2018)
Installation is easiest on a Unix system running Python 3.6. See the Troubleshooting section if you have issues with other configurations.
AIF360 requires specific versions of many Python packages which may conflict with other projects on your system. Virtualenv creates an isolated virtual Python environment where these dependencies may be installed safely. If you have trouble installing AIF360, try this first.
mkdir ~/virtualenvs && cd ~/virtualenvs # this can be wherever you like storing virtualenvs
virtualenv -p python3 aif360 # or substitute your preferred version of Python
source aif360/bin/activateFor Windows, this is a little different:
md C:\virtualenvs # this can be wherever you like storing virtualenvs
cd C:\virtualenvs
virtualenv -p python3 aif360 # or substitute your preferred version of Python
aif360/Scripts/activateThe shell should now look like (aif360) $.
Also, upgrade pip to be safe:
(aif360)$ pip install --upgrade pipTo deactivate the environment, run
deactivateThe prompt will return to $ .
See the Virtualenv User Guide for more details.
pip install aif360This package supports Python 2.7, 3.5, and 3.6. However, for Python 2, the BlackBoxAuditing package must be installed manually.
Clone the latest version of this repository:
git clone https://github.com/IBM/AIF360If you'd like to run the examples, download the datasets now and place them in their respective folders as described in aif360/data/README.md.
Then, navigate to the root directory of the project and run:
pip install .To run the example notebooks, install the additional requirements as follows:
pip install -r requirements.txtThen, follow the Getting Started instructions from PyTorch to download and install the latest version for your machine.
Finally, if you did not already, download the datasets as described in aif360/data/README.md but place them in the appropriate sub-folder in ~/virtualenvs/aif360/lib/pythonX.X/site-packages/aif360/data/raw where pythonX.X is the version of python you are using.
If you encounter any errors during the installation process, look for your issue here and try the solutions.
Note: TensorFlow 1.1.0 only supports Python 3.5 officially on Windows. You can get it from python.org.
Then, try:
pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.1.0-cp35-cp35m-win_amd64.whlTensorFlow is only required for use with the aif360.algorithms.inprocessing.AdversarialDebiasing class.
You may need to download the Visual Studio C++ compiler for Python and rerun
pip install cvxpy==0.4.11CVXPY is only required for use with the aif360.algorithms.preprocessing.OptimPreproc class.
Some additional installation is required to use aif360.algorithms.preprocessing.DisparateImpactRemover with Python 2.7. In a directory of your choosing, run:
git clone https://github.com/algofairness/BlackBoxAuditingIn the root directory of BlackBoxAuditing, run:
echo -n $PWD/BlackBoxAuditing/weka.jar > python2_source/BlackBoxAuditing/model_factories/weka.path
echo "include python2_source/BlackBoxAuditing/model_factories/weka.path" >> MANIFEST.in
pip install --no-deps .This will produce a minimal installation which satisfies our requirements.
The examples directory contains a diverse collection of jupyter notebooks that use AI Fairness 360 in various ways.
Both tutorials and demos illustrate working code using AIF360. Tutorials provide additional discussion that walks the
user through the various steps of the notebook. See the details about tutorials and demos here
A technical description of AI Fairness 360 is available in this paper. Below is the bibtex entry for this paper.
@misc{aif360-oct-2018,
title = "{AI Fairness} 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias",
author = {Rachel K. E. Bellamy and Kuntal Dey and Michael Hind and
Samuel C. Hoffman and Stephanie Houde and Kalapriya Kannan and
Pranay Lohia and Jacquelyn Martino and Sameep Mehta and
Aleksandra Mojsilovic and Seema Nagar and Karthikeyan Natesan Ramamurthy and
John Richards and Diptikalyan Saha and Prasanna Sattigeri and
Moninder Singh and Kush R. Varshney and Yunfeng Zhang},
month = oct,
year = {2018},
url = {https://arxiv.org/abs/1810.01943}
}