(F)airness, (A)ccountability, (T)ransparency and (E)xplainability in Data-driven Decision Systems

Overview Publications Software People

Overview

The concepts of fairness, accountability, transparency and explainability have been studied previously in philosophical, social, cultural, economic and legal frameworks. However, as automated data analysis replaces human supervision and intuition in decision making, and the scale of the data analyzed becomes big, there is a growing need for incorporating these concepts into algorithmic (as well as human) decision-making frameworks. While these concepts are somewhat intuitive, quantifying and incorporating them in algorithmic frameworks is a non-trivial task. For example, what does it mean for an algorithm to be fair or unbiased? Or, how do we make sure that an algorithm does not (un)intentionally discriminate against people from certain social backgrounds? Our work tries to address these issues by proposing novel techniques to detect, quantify and remove such biases from decision making systems.

Publications

• Human Perceptions of Fairness in Algorithmic Decision Making: A Case Study of Criminal Risk Prediction
  by Nina Grgić-Hlača, Elissa M. Redmiles, Krishna P. Gummadi, and Adrian Weller
  To Appear in the Proceedings of the Web Conference (WWW), Lyon, France, April 2018.
  


• Beyond Distributive Fairness in Algorithmic Decision Making: Feature Selection for Procedurally Fair Learning
  by Nina Grgić-Hlača, Muhammad Bilal Zafar, Krishna P. Gummadi, and Adrian Weller
  Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI), New Orleans, LA, February 2018.
  Also presented at the NIPS Symposium on Machine Learning and the Law, Barcelona, Spain, December 2016. (Notable Paper Award)
  


• From Parity to Preference-based Notions of Fairness in Classification
  by Muhammad Bilal Zafar, Isabel Valera, Manuel Gomez-Rodriguez, Krishna P. Gummadi and Adrian Weller
  Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS), Long Beach, CA, December 2017.
  Also presented at the 4th Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT-ML), Halifax, Canada, August 2017.
  


• On Fairness, Diversity and Randomness in Algorithmic Decision Making
  by Nina Grgić-Hlača, Muhammad Bilal Zafar, Krishna P. Gummadi and Adrian Weller
  4th Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT-ML), Halifax, Canada, August 2017.
  


• Who Makes Trends? Understanding Demographic Biases in Crowdsourced Recommendations
  by Abhijnan Chakraborty, Johnnatan Messias, Fabricio Benevenuto, Saptarshi Ghosh, Niloy Ganguly and Krishna P. Gummadi
  Proceedings of the 11th International AAAI Conference on Weblogs and Social Media (ICWSM), Montreal, Canada, May 2017.
  


• Fairness Beyond Disparate Treatment and Disparate Impact: Learning Classification Without Disparate Mistreatment
  by Muhammad Bilal Zafar, Isabel Valera, Manuel Gomez-Rodriguez and Krishna P. Gummadi
  Proceedings of the 26th International World Wide Web Conference (WWW), Perth, Australia, April 2017. (Best Paper Honorable Mention)
  Also presented at the 3rd Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT-ML), New York City, NY, November 2016.
  


• Fairness Constraints: Mechanisms for Fair Classification
  by Muhammad Bilal Zafar, Isabel Valera, Manuel Gomez-Rodriguez and Krishna P. Gummadi
  Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS), Fort Lauderdale, FL, April 2017.
  Also presented at the 2nd Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT-ML), Lille, France, July 2015.
  


• Quantifying Search Bias: Investigating Sources of Bias for Political Searches in Social Media
  by Juhi Kulshrestha, Motahhare Eslami, Johnnatan Messias, Muhammad Bilal Zafar, Saptarshi Ghosh, Krishna P. Gummadi, and Karrie G. Karahalios
  Proceedings of the 20th ACM Conference on Computer Supported Cooperative Work and Social Computing (CSCW), Portland, OR, February 2017.
  


• The Case for Temporal Transparency: Detecting Policy Change Events in Black-Box Decision Making Systems
  by Miguel Ferreira, Muhammad Bilal Zafar, Isabel Valera, and Krishna P. Gummadi
  3rd Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT-ML), New York City, NY, November 2016.
  


• Message Impartiality in Social Media Discussions
  by Muhammad Bilal Zafar, Krishna P. Gummadi, and Cristian Danescu-Niculescu-Mizil
  Proceedings of the 10th International AAAI Conference on Weblogs and Social Media (ICWSM), Cologne, Germany, May 2016.
  


• Can Trending News Stories Create Coverage Bias? On the Impact of High Content Churn in Online News Media
  by Abhijnan Chakraborty, Saptarshi Ghosh, Niloy Ganguly and Krishna P. Gummadi
  Computation + Journalism Symposium, New York, NY, October 2015.
  

Software

• Fairness in Classification
  Code related to our AISTATS'17, WWW'17 and NIPS'17 papers. Provides a Python implementation of fair logistic regression classifiers.


• Search Political Leaning of Twitter Users
  Twitter application where users can log in to see their inferred political leaning (between Democratic- and Republican- leaning) , as well as, search for other Twitter users to check out their inferred political leanings.


• Explore Bias of Social Media News
  Systems to expose the demographic biases of news publishers and promoters on social media.


• Procedural Fairness in Machine Learning
  Code related to our AAAI 2018 and WWW 2018 papers. Provides Python implementations of the submodular optimization framework from our AAAI 2018 paper and code for reproducing the analysis presented in our WWW 2018 paper.

People

Faculty		Student / Post doc
Manuel Gomez Rodriguez Krishna P. Gummadi		Nina Grgić-Hlača Juhi Kulshrestha Till Speicher Muhammad Bilal Zafar Abhijnan Chakraborty

Imprint | Data Protection