Causal Modeling for Fairness in Dynamical Systems

In many application areas---lending, education, and online recommenders, for example---fairness and equity concerns emerge when a machine learning system interacts with a dynamically changing environment to produce both immediate and long-term effects for individuals and demographic groups. We discuss causal directed acyclic graphs (DAGs) as a unifying framework for the recent literature on fairness in such dynamical systems. We show that this formulation affords several new directions of inquiry to the modeler, where causal assumptions can be expressed and manipulated. We emphasize the importance of computing interventional quantities in the dynamical fairness setting, and show how causal assumptions enable simulation (when environment dynamics are known) and off-policy estimation (when dynamics are unknown) of intervention on short- and long-term outcomes, at both the group and individual levels.

[1]  Eric B. Laber,et al.  A Robust Method for Estimating Optimal Treatment Regimes , 2012, Biometrics.

[2]  David F. Larcker,et al.  Structural Equation Models with Unobservable Variables and Measurement Error: Algebra and Statistics: , 1981 .

[3]  D. Rubin Causal Inference Using Potential Outcomes , 2005 .

[4]  J. Pearl Causal inference in statistics: An overview , 2009 .

[5]  Julius Lieblein,et al.  SOME APPLICATIONS OF EXTREME- VALUE METHODS , 1954 .

[6]  A. Dawid Influence Diagrams for Causal Modelling and Inference , 2002 .

[7]  Percy Liang,et al.  Fairness Without Demographics in Repeated Loss Minimization , 2018, ICML.

[8]  Suresh Venkatasubramanian,et al.  Runaway Feedback Loops in Predictive Policing , 2017, FAT.

[9]  R. Avery,et al.  Credit Scoring and Its Effects on the Availability and Affordability of Credit , 2009 .

[10]  David Silver,et al.  Credit Assignment Techniques in Stochastic Computation Graphs , 2019, AISTATS.

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

[12]  Petra Wächter Transitions to sustainable development – new directions in the study of long term transformative change , 2012 .

[13]  David A. Scanlan Structured flowcharts outperform pseudocode: an experimental comparison , 1989, IEEE Software.

[14]  Alexander D'Amour,et al.  Fairness is not static: deeper understanding of long term fairness via simulation studies , 2020, FAT*.

[15]  R. Kessler,et al.  Widowhood and depression: explaining long-term gender differences in vulnerability. , 1992, Journal of health and social behavior.

[16]  Nir Friedman,et al.  Probabilistic Graphical Models - Principles and Techniques , 2009 .

[17]  Wayne A. Fuller,et al.  Measurement Error Models , 1988 .

[18]  Nicolas Heess,et al.  Woulda, Coulda, Shoulda: Counterfactually-Guided Policy Search , 2018, ICLR.

[19]  Peter Schwartz,et al.  The art of the long view : paths to strategic insight for yourself and your company , 1996 .

[20]  David Sontag,et al.  Counterfactual Off-Policy Evaluation with Gumbel-Max Structural Causal Models , 2019, ICML.

[21]  Ilya Shpitser,et al.  Learning Optimal Fair Policies , 2018, ICML.

[22]  Richard Kammann,et al.  The Comprehensibility of Printed Instructions and the Flowchart Alternative , 1975 .

[23]  Toniann Pitassi,et al.  Fairness through Causal Awareness: Learning Causal Latent-Variable Models for Biased Data , 2018, FAT.

[24]  Judea Pearl,et al.  On the Consistency Rule in Causal Inference: Axiom, Definition, Assumption, or Theorem? , 2010, Epidemiology.

[25]  Anca D. Dragan,et al.  The Social Cost of Strategic Classification , 2018, FAT.

[26]  J. Robins,et al.  Doubly Robust Estimation in Missing Data and Causal Inference Models , 2005, Biometrics.

[27]  Brad A. Myers,et al.  Taxonomies of visual programming and program visualization , 1990, J. Vis. Lang. Comput..

[28]  Philip S. Thomas,et al.  Importance Sampling for Fair Policy Selection , 2017, UAI.

[29]  Nathan Srebro,et al.  From Fair Decision Making To Social Equality , 2018, FAT.

[30]  Lois M. Haibt,et al.  A program to draw multilevel flow charts , 1899, IRE-AIEE-ACM '59 (Western).

[31]  T. Richardson Single World Intervention Graphs ( SWIGs ) : A Unification of the Counterfactual and Graphical Approaches to Causality , 2013 .

[32]  Donald E. Knuth,et al.  Computer-drawn flowcharts , 1963, CACM.

[33]  John Langford,et al.  Doubly Robust Policy Evaluation and Learning , 2011, ICML.

[34]  Elias Bareinboim,et al.  Fairness in Decision-Making - The Causal Explanation Formula , 2018, AAAI.

[35]  Sebastian Benthall,et al.  Racial categories in machine learning , 2018, FAT.

[36]  Aaron Roth,et al.  Fairness in Reinforcement Learning , 2016, ICML.

[37]  M. Gribaudo,et al.  2002 , 2001, Cell and Tissue Research.

[38]  R. Mayer Different problem-solving competencies established in learning computer programming with and without meaningful models. , 1975 .

[39]  Dean P. Foster,et al.  An Economic Argument for Affirmative Action , 1992 .

[40]  M. G. Morgan,et al.  Why Conventional Tools for Policy Analysis Are Often Inadequate for Problems of Global Change , 1999 .

[41]  Shane Legg,et al.  Understanding Agent Incentives using Causal Influence Diagrams. Part I: Single Action Settings , 2019, ArXiv.

[42]  Nan Jiang,et al.  Doubly Robust Off-policy Value Evaluation for Reinforcement Learning , 2015, ICML.

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

[44]  Ilya Shpitser,et al.  Fair Inference on Outcomes , 2017, AAAI.

[45]  A. Satorra,et al.  Measurement Error Models , 1988 .

[46]  Nava Tintarev,et al.  SIREN: A Simulation Framework for Understanding the Effects of Recommender Systems in Online News Environments , 2019, FAT.

[47]  Philip S. Thomas,et al.  Data-Efficient Off-Policy Policy Evaluation for Reinforcement Learning , 2016, ICML.

[48]  Nicole Immorlica,et al.  The Disparate Effects of Strategic Manipulation , 2018, FAT.

[49]  Bridget J. Goosby,et al.  The Transgenerational Consequences of Discrimination on African-American Health Outcomes: Discrimination and Health , 2013 .

[50]  Pieter Abbeel,et al.  Gradient Estimation Using Stochastic Computation Graphs , 2015, NIPS.

[51]  Sampath Kannan,et al.  Downstream Effects of Affirmative Action , 2018, FAT.

[52]  Stephen Coate,et al.  Will Affirmative-Action Policies Eliminate Negative Stereotypes? , 1993 .

[53]  Emily Denton,et al.  Towards a critical race methodology in algorithmic fairness , 2019, FAT*.

[54]  Tom Minka,et al.  A* Sampling , 2014, NIPS.

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

[56]  Yiling Chen,et al.  A Short-term Intervention for Long-term Fairness in the Labor Market , 2017, WWW.

[57]  Doina Precup,et al.  Eligibility Traces for Off-Policy Policy Evaluation , 2000, ICML.

[58]  Aaron Roth,et al.  Fairness in Learning: Classic and Contextual Bandits , 2016, NIPS.

[59]  K. Lum,et al.  To predict and serve? , 2016 .

[60]  David G. Groves,et al.  A General, Analytic Method for Generating Robust Strategies and Narrative Scenarios , 2006, Manag. Sci..

[61]  Matt J. Kusner,et al.  Making Decisions that Reduce Discriminatory Impacts , 2019, ICML.

[62]  Paul Goldsmith-Pinkham,et al.  Predictably Unequal? The Effects of Machine Learning on Credit Markets , 2017, The Journal of Finance.

[63]  P. Wright,et al.  Written information: Some alternatives to prose for expressing the outcomes of complex contingencies. , 1973 .

[64]  Elias Bareinboim,et al.  Equality of Opportunity in Classification: A Causal Approach , 2018, NeurIPS.

[65]  Guido W. Imbens,et al.  Potential Outcome and Directed Acyclic Graph Approaches to Causality: Relevance for Empirical Practice in Economics , 2019, Journal of Economic Literature.

[66]  Saltelli Andrea,et al.  Global Sensitivity Analysis: The Primer , 2008 .

[67]  Susan Athey,et al.  Machine Learning and Causal Inference for Policy Evaluation , 2015, KDD.

[68]  John Langford,et al.  Off-policy evaluation for slate recommendation , 2016, NIPS.

[69]  Herman H. Goldstine,et al.  Planning and coding of problems for an Electronic Computing Instrument , 1947 .

[70]  James J. Heckman,et al.  Econometric Evaluation of Social Programs, Part I: Causal Models, Structural Models and Econometric Policy Evaluation , 2007 .

[71]  Esther Rolf,et al.  Delayed Impact of Fair Machine Learning , 2018, ICML.

[72]  Paul R. Rosenbaum,et al.  Sensitivity Analysis in Observational Studies , 2005 .