An Information-Theoretic Quantification of Discrimination with Exempt Features

The needs of a business (e.g., hiring) may require the use of certain features that are critical in a way that any discrimination arising due to them should be exempted. In this work, we propose a novel information-theoretic decomposition of the total discrimination (in a counterfactual sense) into a non-exempt component, which quantifies the part of the discrimination that cannot be accounted for by the critical features, and an exempt component, which quantifies the remaining discrimination. Our decomposition enables selective removal of the non-exempt component if desired. We arrive at this decomposition through examples and counterexamples that enable us to first obtain a set of desirable properties that any measure of non-exempt discrimination should satisfy. We then demonstrate that our proposed quantification of non-exempt discrimination satisfies all of them. This decomposition leverages a body of work from information theory called Partial Information Decomposition (PID). We also obtain an impossibility result showing that no observational measure of non-exempt discrimination can satisfy all of the desired properties, which leads us to relax our goals and examine alternative observational measures that satisfy only some of these properties. We then perform a case study using one observational measure to show how one might train a model allowing for exemption of discrimination due to critical features.

[1]  Krishna P. Gummadi,et al.  Fairness Constraints: Mechanisms for Fair Classification , 2015, AISTATS.

[2]  Bernhard Schölkopf,et al.  Avoiding Discrimination through Causal Reasoning , 2017, NIPS.

[3]  Junpei Komiyama,et al.  Two-stage Algorithm for Fairness-aware Machine Learning , 2017, ArXiv.

[4]  Guido Montúfar,et al.  Computing the Unique Information , 2017, 2018 IEEE International Symposium on Information Theory (ISIT).

[5]  Dan Suciu,et al.  Interventional Fairness: Causal Database Repair for Algorithmic Fairness , 2019, SIGMOD Conference.

[6]  Murray Shanahan,et al.  The Partial Information Decomposition of Generative Neural Network Models , 2017, Entropy.

[7]  Eckehard Olbrich,et al.  Quantifying unique information , 2013, Entropy.

[8]  Toniann Pitassi,et al.  Fairness through awareness , 2011, ITCS '12.

[9]  Bernhard Schölkopf,et al.  Elements of Causal Inference: Foundations and Learning Algorithms , 2017 .

[10]  G. Crooks.  On Measures of Entropy and Information , 2015 .

[11]  Nathan Srebro,et al.  Equality of Opportunity in Supervised Learning , 2016, NIPS.

[12]  John Langford,et al.  A Reductions Approach to Fair Classification , 2018, ICML.

[13]  K. Manley,et al.  The BFOQ Defense: Title VII's Concession to Gender Discrimination , 2009 .

[14]  Berk Ustun,et al.  Repairing without Retraining: Avoiding Disparate Impact with Counterfactual Distributions , 2019, ICML.

[15]  AmirEmad Ghassami,et al.  Fairness in Supervised Learning: An Information Theoretic Approach , 2018, 2018 IEEE International Symposium on Information Theory (ISIT).

[16]  Panagiotis Papapetrou,et al.  A peek into the black box: exploring classifiers by randomization , 2014, Data Mining and Knowledge Discovery.

[17]  Kush R. Varshney,et al.  Optimized Pre-Processing for Discrimination Prevention , 2017, NIPS.

[18]  Matt J. Kusner,et al.  When Worlds Collide: Integrating Different Counterfactual Assumptions in Fairness , 2017, NIPS.

[19]  Faisal Kamiran,et al.  Quantifying explainable discrimination and removing illegal discrimination in automated decision making , 2012, Knowledge and Information Systems.

[20]  Toniann Pitassi,et al.  Learning Fair Representations , 2013, ICML.

[21]  Matt J. Kusner,et al.  Counterfactual Fairness , 2017, NIPS.

[22]  Percy Liang,et al.  Understanding Black-box Predictions via Influence Functions , 2017, ICML.

[23]  Sang Joon Kim,et al.  A Mathematical Theory of Communication , 2006 .

[24]  Suresh Venkatasubramanian,et al.  Auditing black-box models for indirect influence , 2016, Knowledge and Information Systems.

[25]  Pulkit Grover,et al.  Information Flow in Computational Systems , 2019, IEEE Transactions on Information Theory.

[26]  Avi Feller,et al.  Algorithmic Decision Making and the Cost of Fairness , 2017, KDD.

[27]  Jun Sakuma,et al.  Fairness-Aware Classifier with Prejudice Remover Regularizer , 2012, ECML/PKDD.

[28]  Eckehard Olbrich,et al.  Unique Information and Secret Key Decompositions , 2019, 2019 IEEE International Symposium on Information Theory (ISIT).

[29]  Yair Zick,et al.  Algorithmic Transparency via Quantitative Input Influence: Theory and Experiments with Learning Systems , 2016, 2016 IEEE Symposium on Security and Privacy (SP).

[30]  Silvia Chiappa,et al.  Path-Specific Counterfactual Fairness , 2018, AAAI.

[31]  Krishna P. Gummadi,et al.  Fairness Behind a Veil of Ignorance: A Welfare Analysis for Automated Decision Making , 2018, NeurIPS.

[32]  Shai Ben-David,et al.  Empirical Risk Minimization under Fairness Constraints , 2018, NeurIPS.

[33]  Peter Kairouz,et al.  Learning Generative Adversarial RePresentations (GAP) under Fairness and Censoring Constraints , 2019, ArXiv.

[34]  Matt Fredrikson,et al.  Use Privacy in Data-Driven Systems: Theory and Experiments with Machine Learnt Programs , 2017, CCS.

[35]  Andrew D. Selbst,et al.  Big Data's Disparate Impact , 2016 .

[36]  Aditya Krishna Menon,et al.  The cost of fairness in binary classification , 2018, FAT.

[37]  Kush R. Varshney,et al.  Trustworthy machine learning and artificial intelligence , 2019, XRDS.

[38]  Randall D. Beer,et al.  Nonnegative Decomposition of Multivariate Information , 2010, ArXiv.