ABSTRACT
Systems that offer continuous model monitoring have emerged in response to (1) well-documented failures of deployed Machine Learning (ML) and Artificial Intelligence (AI) models and (2) new regulatory requirements impacting these models. Existing monitoring systems continuously track the performance of deployed ML models and compute feature importance (a.k.a. explanations) for each prediction to help developers identify the root causes of emergent model performance problems.
We present Quantile Demographic Drift (QDD), a novel model bias quantification metric that uses quantile binning to measure differences in the overall prediction distributions over subgroups. QDD is ideal for continuous monitoring scenarios, does not suffer from the statistical limitations of conventional threshold-based bias metrics, and does not require outcome labels (which may not be available at runtime). We incorporate QDD into a continuous model monitoring system, called FairCanary, that reuses existing explanations computed for each individual prediction to quickly compute explanations for the QDD bias metrics. This optimization makes FairCanary an order of magnitude faster than previous work that has tried to generate feature-level bias explanations.
Supplemental Material
- 116th Congress (2019--2020). [n.,d.]. H.R.2231 - Algorithmic Accountability Act of 2019. https://www.congress.gov/bill/116th-congress/house-bill/2231.Google Scholar
- Solon Barocas, Anhong Guo, Ece Kamar, Jacquelyn Krones, Meredith Ringel Morris, Jennifer Wortman Vaughan, Duncan Wadsworth, and Hanna Wallach. 2021. Designing Disaggregated Evaluations of AI Systems: Choices, Considerations, and Tradeoffs. arXiv preprint arXiv:2103.06076 (2021).Google Scholar
- Rachel KE Bellamy, Kuntal Dey, Michael Hind, Samuel C Hoffman, Stephanie Houde, Kalapriya Kannan, Pranay Lohia, Jacquelyn Martino, Sameep Mehta, Aleksandra Mojsilovic, et al. 2018. AI Fairness 360: An extensible toolkit for detecting, understanding, and mitigating unwanted algorithmic bias. arXiv preprint arXiv:1810.01943 (2018).Google Scholar
- Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. 2021. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency.Google ScholarDigital Library
- Umang Bhatt, Alice Xiang, Shubham Sharma, Adrian Weller, Ankur Taly, Yunhan Jia, Joydeep Ghosh, Ruchir Puri, José MF Moura, and Peter Eckersley. 2020. Explainable machine learning in deployment. In Proceedings of the 2020 conference on fairness, accountability, and transparency. 648--657.Google ScholarDigital Library
- Albert Bifet and Ricard Gavalda. 2007. Learning from time-changing data with adaptive windowing. In Proceedings of the 2007 SIAM international conference on data mining. SIAM, 443--448.Google ScholarCross Ref
- Alexander Binder, Grégoire Montavon, Sebastian Lapuschkin, Klaus-Robert Müller, and Wojciech Samek. 2016. Layer-wise relevance propagation for neural networks with local renormalization layers. In International Conference on Artificial Neural Networks. Springer, 63--71.Google ScholarCross Ref
- Emily Black, Samuel Yeom, and Matt Fredrikson. 2020. Fliptest: fairness testing via optimal transport. In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency. 111--121.Google ScholarDigital Library
- Miranda Bogen and Aaron Rieke. 2018. Help wanted: An examination of hiring algorithms, equity, and bias. (2018).Google Scholar
- Eric Breck, Neoklis Polyzotis, Sudip Roy, Steven Whang, and Martin Zinkevich. 2019. Data Validation for Machine Learning.. In MLSys.Google Scholar
- European Commission. [n.,d.]. Proposal for a Regulation laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). https://digital-strategy.ec.europa.eu/en/library/proposal-regulation-laying-down-harmonised-rules-artificial-intelligence-artificial-intelligence.Google Scholar
- Sam Corbett-Davies and Sharad Goel. 2018. The measure and mismeasure of fairness: A critical review of fair machine learning. arXiv preprint arXiv:1808.00023 (2018).Google Scholar
- Sam Corbett-Davies, Emma Pierson, Avi Feller, Sharad Goel, and Aziz Huq. 2017. Algorithmic decision making and the cost of fairness. In Proceedings of the 23rd acm sigkdd international conference on knowledge discovery and data mining. 797--806.Google ScholarDigital Library
- Jakub Czakon. 2022. Best Tools to Do ML Model Monitoring. (2022). https://neptune.ai/blog/ml-model-monitoring-best-toolsGoogle Scholar
- Sanjiv Das, Michele Donini, Jason Gelman, Kevin Haas, Mila Hardt, Jared Katzman, Krishnaram Kenthapadi, Pedro Larroy, Pinar Yilmaz, and Muhammad Bilal Zafar. 2021. Fairness Measures for Machine Learning in Finance. The Journal of Financial Data Science, Vol. 3, 4 (2021), 33--64.Google ScholarCross Ref
- Anupam Datta, Shayak Sen, and Yair Zick. 2016. Algorithmic transparency via quantitative input influence: Theory and experiments with learning systems. In 2016 IEEE symposium on security and privacy (SP). IEEE, 598--617.Google ScholarCross Ref
- Denis Moreira dos Reis, Peter Flach, Stan Matwin, and Gustavo Batista. 2016. Fast unsupervised online drift detection using incremental kolmogorov-smirnov test. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 1545--1554.Google ScholarDigital Library
- Cynthia Dwork, Moritz Hardt, Toniann Pitassi, Omer Reingold, and Richard Zemel. 2012. Fairness through awareness. In Proceedings of the 3rd innovations in theoretical computer science conference. 214--226.Google ScholarDigital Library
- Equal Employment Opportunity Commission, Civil Service Commission, et al. 1978. Uniform guidelines on employee selection procedures. Federal Register, Vol. 43, 166 (1978), 38290--38315.Google Scholar
- UK Office for Artificial Intelligence. [n.,d.]. Ethics, Transparency and Accountability Framework for Automated Decision-Making. https://www.gov.uk/government/publications/ethics-transparency-and-accountability-framework-for-automated-decision-making.Google Scholar
- Center for Data Science and Public Policy. [n.,d.]. Aequitas: Fairness Tree. http://www.datasciencepublicpolicy.org/projects/aequitas/.Google Scholar
- Joao Gama, Raquel Sebastiao, and Pedro Pereira Rodrigues. 2013. On evaluating stream learning algorithms. Machine learning, Vol. 90, 3 (2013), 317--346.Google Scholar
- Sahin Cem Geyik, Stuart Ambler, and Krishnaram Kenthapadi. 2019. Fairness-aware ranking in search & recommendation systems with application to linkedin talent search. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2221--2231.Google ScholarDigital Library
- Sindhu Ghanta, Sriram Subramanian, Lior Khermosh, Swaminathan Sundararaman, Harshil Shah, Yakov Goldberg, Drew S. Roselli, and Nisha Talagala. 2019. ML Health: Fitness Tracking for Production Models. CoRR, Vol. abs/1902.02808 (2019). arxiv: 1902.02808 http://arxiv.org/abs/1902.02808Google Scholar
- Avijit Ghosh, Lea Genuit, and Mary Reagan. 2021. Characterizing Intersectional Group Fairness with Worst-Case Comparisons. arXiv preprint arXiv:2101.01673 (2021).Google Scholar
- Vincent Grari, Boris Ruf, Sylvain Lamprier, and Marcin Detyniecki. 2019. Fairness-Aware Neural Réyni Minimization for Continuous Features. arXiv preprint arXiv:1911.04929 (2019).Google Scholar
- Moritz Hardt, Eric Price, and Nathan Srebro. 2016. Equality of opportunity in supervised learning. arXiv preprint arXiv:1610.02413 (2016).Google Scholar
- Sara Hooker, Dumitru Erhan, Pieter-Jan Kindermans, and Been Kim. 2019. A benchmark for interpretability methods in deep neural networks. Advances in neural information processing systems, Vol. 32 (2019).Google Scholar
- Ray Jiang, Aldo Pacchiano, Tom Stepleton, Heinrich Jiang, and Silvia Chiappa. 2020. Wasserstein fair classification. In Uncertainty in Artificial Intelligence. PMLR, 862--872.Google Scholar
- Faisal Kamiran and Toon Calders. 2012. Data preprocessing techniques for classification without discrimination. Knowledge and information systems, Vol. 33, 1 (2012), 1--33.Google Scholar
- Amir-Hossein Karimi, Gilles Barthe, Bernhard Schölkopf, and Isabel Valera. 2020. A survey of algorithmic recourse: definitions, formulations, solutions, and prospects. arXiv preprint arXiv:2010.04050 (2020).Google Scholar
- Niki Kilbertus, Philip J Ball, Matt J Kusner, Adrian Weller, and Ricardo Silva. 2020. The sensitivity of counterfactual fairness to unmeasured confounding. In Uncertainty in artificial intelligence. PMLR, 616--626.Google Scholar
- Jon Kleinberg, Sendhil Mullainathan, and Manish Raghavan. 2016. Inherent trade-offs in the fair determination of risk scores. arXiv preprint arXiv:1609.05807 (2016).Google Scholar
- Alistair Knott. [n.,d.]. Moving Towards Responsible Government Use of AI in New Zealand). https://digitaltechitp.nz/2021/03/22/moving-towards-responsible-government-use-of-ai-in-new-zealand/.Google Scholar
- I Elizabeth Kumar, Suresh Venkatasubramanian, Carlos Scheidegger, and Sorelle Friedler. 2020 a. Problems with Shapley-value-based explanations as feature importance measures. In International Conference on Machine Learning. PMLR, 5491--5500.Google Scholar
- I. Elizabeth Kumar, Suresh Venkatasubramanian, Carlos Scheidegger, and Sorelle Friedler. 2020 b. Problems with Shapley-value-based explanations as feature importance measures. In Proceedings of the 37th International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 119), Hal Daumé III and Aarti Singh (Eds.). PMLR, 5491--5500. https://proceedings.mlr.press/v119/kumar20e.htmlGoogle Scholar
- Matt J Kusner, Joshua R Loftus, Chris Russell, and Ricardo Silva. 2017. Counterfactual fairness. arXiv preprint arXiv:1703.06856 (2017).Google Scholar
- Jianhua Lin. 1991. Divergence measures based on the Shannon entropy. IEEE Transactions on Information theory, Vol. 37, 1 (1991), 145--151.Google ScholarDigital Library
- Lydia T Liu, Sarah Dean, Esther Rolf, Max Simchowitz, and Moritz Hardt. 2018. Delayed impact of fair machine learning. In International Conference on Machine Learning. PMLR, 3150--3158.Google Scholar
- Scott M Lundberg and Su-In Lee. 2017. A Unified Approach to Interpreting Model Predictions. In Advances in Neural Information Processing Systems 30, I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (Eds.). Curran Associates, Inc., 4765--4774. http://papers.nips.cc/paper/7062-a-unified-approach-to-interpreting-model-predictions.pdfGoogle Scholar
- Aniek F Markus, Jan A Kors, and Peter R Rijnbeek. 2021. The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies. Journal of Biomedical Informatics, Vol. 113 (2021), 103655.Google ScholarDigital Library
- Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. 2019. A survey on bias and fairness in machine learning. arXiv preprint arXiv:1908.09635 (2019).Google Scholar
- Luke Merrick and Ankur Taly. 2020. The Explanation Game: Explaining Machine Learning Models Using Shapley Values. In International Cross-Domain Conference for Machine Learning and Knowledge Extraction. Springer, 17--38.Google Scholar
- Alexey Miroshnikov, Konstandinos Kotsiopoulos, Ryan Franks, and Arjun Ravi Kannan. 2020. Wasserstein-based fairness interpretability framework for machine learning models. arXiv preprint arXiv:2011.03156 (2020).Google Scholar
- Preetam Nandy, Cyrus Diciccio, Divya Venugopalan, Heloise Logan, Kinjal Basu, and Noureddine El Karoui. 2021. Achieving Fairness via Post-Processing in Web-Scale Recommender Systems. arxiv: 2006.11350 [stat.ML]Google Scholar
- Arvind Narayanan. [n.,d.]. 21 fairness definitions and their politics. https://fairmlbook.org/tutorial2.html.Google Scholar
- David Nigenda, Zohar Karnin, Muhammad Bilal Zafar, Raghu Ramesha, Alan Tan, Michele Donini, and Krishnaram Kenthapadi. 2021. Amazon SageMaker Model Monitor: A System for Real-Time Insights into Deployed Machine Learning Models. arXiv preprint arXiv:2111.13657 (2021).Google Scholar
- K. Nishida, S. Shimada, S. Ishikawa, and K. Yamauchi. 2008. Detecting sudden concept drift with knowledge of human behavior. In 2008 IEEE International Conference on Systems, Man and Cybernetics. 3261--3267. https://doi.org/10.1109/ICSMC.2008.4811799Google ScholarCross Ref
- Government of Canada. [n.,d.]. Responsible use of artificial intelligence (AI). https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai.html.Google Scholar
- Fábio Pinto, Marco OP Sampaio, and Pedro Bizarro. 2019. Automatic model monitoring for data streams. arXiv preprint arXiv:1908.04240 (2019).Google Scholar
- Manish Raghavan, Solon Barocas, Jon Kleinberg, and Karen Levy. 2020. Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practices. In Proceedings of the ACM Conference on Fairness, Accountability, and Transparency (FAT*).Google ScholarDigital Library
- Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. "Why should i trust you?" Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. 1135--1144.Google ScholarDigital Library
- Marcos Salganicoff. 1997. Tolerating concept and sampling shift in lazy learning using prediction error context switching. In Lazy learning. Springer, 133--155.Google Scholar
- Sebastian Schelter, Felix Biessmann, Tim Januschowski, David Salinas, Stephan Seufert, and Gyuri Szarvas. 2018. On challenges in machine learning model management. (2018).Google Scholar
- David Sculley, Gary Holt, Daniel Golovin, Eugene Davydov, Todd Phillips, Dietmar Ebner, Vinay Chaudhary, Michael Young, Jean-Francois Crespo, and Dan Dennison. 2015. Hidden technical debt in machine learning systems. Advances in neural information processing systems, Vol. 28 (2015).Google Scholar
- Andrew Selbst and Julia Powles. 2018. "Meaningful Information" and the Right to Explanation. In Conference on Fairness, Accountability and Transparency. PMLR, 48--48.Google Scholar
- Andrew D Selbst, Danah Boyd, Sorelle A Friedler, Suresh Venkatasubramanian, and Janet Vertesi. 2019. Fairness and abstraction in sociotechnical systems. In Proceedings of the conference on fairness, accountability, and transparency. 59--68.Google ScholarDigital Library
- Aalok Shanbhag, Avijit Ghosh, and Josh Rubin. 2021. Unified Shapley Framework to Explain Prediction Drift. arXiv preprint arXiv:2102.07862 (2021).Google Scholar
- Avanti Shrikumar, Peyton Greenside, and Anshul Kundaje. 2017. Learning important features through propagating activation differences. In International conference on machine learning. PMLR, 3145--3153.Google Scholar
- Kenneth O Stanley. 2003. Learning concept drift with a committee of decision trees. Informe técnico: UT-AI-TR-03-302, Department of Computer Sciences, University of Texas at Austin, USA (2003).Google Scholar
- Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In International Conference on Machine Learning. PMLR, 3319--3328.Google Scholar
- Cédric Villani. 2009. The wasserstein distances. In Optimal transport. Springer, 93--111.Google Scholar
- Christo Wilson, Avijit Ghosh, Shan Jiang, Alan Mislove, Lewis Baker, Janelle Szary, Kelly Trindel, and Frida Polli. 2021. Building and auditing fair algorithms: A case study in candidate screening. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. 666--677.Google ScholarDigital Library
- Indre vZ liobaite. 2010. Change with delayed labeling: When is it detectable?. In 2010 IEEE International Conference on Data Mining Workshops. IEEE, 843--850.Google Scholar
Index Terms
- FairCanary: Rapid Continuous Explainable Fairness
Recommendations
PreCoF: counterfactual explanations for fairness
AbstractThis paper studies how counterfactual explanations can be used to assess the fairness of a model. Using machine learning for high-stakes decisions is a threat to fairness as these models can amplify bias present in the dataset, and there is no ...
Graphical Perception of Saliency-based Model Explanations
CHI '23: Proceedings of the 2023 CHI Conference on Human Factors in Computing SystemsIn recent years, considerable work has been devoted to explaining predictive, deep learning-based models, and in turn how to evaluate explanations. An important class of evaluation methods are ones that are human-centered, which typically require the ...
Bias in Artificial Intelligence Models in Financial Services
AIES '22: Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and SocietyNowadays, artificial intelligence models are widely used in financial services, from credit scoring to fraud detection, having a direct impact on our daily lives. Although such models have been developed to try to reduce human bias and thus bring ...
Comments