FairML : ToolBox for diagnosing bias in predictive modeling

By Julius Adebayo. 2016

101 citations recorded by the source corpus.

Subfields: Evaluation & Benchmarks, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Evaluation & Benchmarks

1. Jesse London
2. Loren Hinkson
3. Dena F. Mujtaba
4. Julien Ferry
5. Prasanna Sattigeri
6. Kush R. Varshney
7. Sébastien Gambs
8. Rayid Ghani

In AI Ethics, Fairness & Society

1. Jesse London
2. Loren Hinkson
3. Aleksandra Mojsilovic
4. Dena F. Mujtaba
5. Julien Ferry
6. Karthikeyan Natesan Ramamurthy
7. Kush R. Varshney
8. Sébastien Gambs

Citations in the corpus

• Aequitas: A Bias and Fairness Audit Toolkit (2018, methodology): “FairML Adebayo (2016) is a library that can be categorized as an attempt to increase interpretability of the predictions of black box models with respect to perturbations in the inputs, and therefore can also be used to measure how different models react to data”
• AI Fairness 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias (2018, methodology): “Similarly, FairML (Adebayo, 2016) provides an auditing tool for predictive models by quantifying the relative effects of various inputs on a models predictions.”
• Learning Fair Rule Lists (2019, methodology): “We used the FairML [1] library to audit the models we produced.”
• Ethical Considerations in AI-Based Recruitment (2019, methodology): “ FairML [35]: FairML focuses on assisting in the postprocessing stage of bias mitigation, by explaining the significance of a model’s features.”