FairCanary: Rapid Continuous Explainable Fairness

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.

arXiv commentAccepted as a full paper at the 2022 AAAI/ACM Conference on AI, Ethics, and Society

References in corpus (17)

• A Unified Approach to Interpreting Model Predictions
• Axiomatic Attribution for Deep Networks
• Learning Important Features Through Propagating Activation Differences
• Equality of Opportunity in Supervised Learning
• Counterfactual Fairness
• The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies
• Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practices
• The Measure and Mismeasure of Fairness
• A Benchmark for Interpretability Methods in Deep Neural Networks
• Fairness-Aware Ranking in Search & Recommendation Systems with Application to LinkedIn Talent Search
• FlipTest: Fairness Testing via Optimal Transport
• The Sensitivity of Counterfactual Fairness to Unmeasured Confounding
• Wasserstein-based fairness interpretability framework for machine learning models
• Fairness-Aware Neural Réyni Minimization for Continuous Features
• Automatic Model Monitoring for Data Streams
• ML Health: Fitness Tracking for Production Models
• Unified Shapley Framework to Explain Prediction Drift

Cited by in corpus (5)

• Impossibility Theorems for Feature Attribution
• A Critical Survey on Fairness Benefits of Explainable AI
• Toward Operationalizing Pipeline-aware ML Fairness: A Research Agenda for Developing Practical Guidelines and Tools
• Disentangling and Operationalizing AI Fairness at LinkedIn
• When Fair Classification Meets Noisy Protected Attributes