Achieving Group Fairness through Independence in Predictive Process Monitoring

2024-12-06

arXiv.org

AI Summary

This work addresses group unfairness in predictive process monitoring arising from historical data bias. To mitigate this, we propose a fairness-aware method grounded in statistical independence. Our core contribution is the first incorporation of the Wasserstein distance into process monitoring, enabling a tunable composite loss function that jointly optimizes prediction accuracy (via binary cross-entropy) and group fairness (via distribution alignment). We further adopt threshold-free fairness metrics—including ΔDP and distributional surrogates—for robust fairness evaluation. In controlled experiments, our approach reduces group bias (ΔDP) by 62% while preserving over 90% of the original predictive performance. This establishes a verifiable, tunable paradigm for algorithmic fairness in process monitoring—bridging predictive utility and equitable decision-making without reliance on arbitrary classification thresholds.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Bias, Fairness & Privacy
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

Predictive process monitoring focuses on forecasting future states of ongoing process executions, such as predicting the outcome of a particular case. In recent years, the application of machine learning models in this domain has garnered significant scientific attention. When using historical execution data, which may contain biases or exhibit unfair behavior, these biases may be encoded into the trained models. Consequently, when such models are deployed to make decisions or guide interventions for new cases, they risk perpetuating this unwanted behavior. This work addresses group fairness in predictive process monitoring by investigating independence, i.e. ensuring predictions are unaffected by sensitive group membership. We explore independence through metrics for demographic parity such as $Delta$DP, as well as recently introduced, threshold-independent distribution-based alternatives. Additionally, we propose a composite loss function existing of binary cross-entropy and a distribution-based loss (Wasserstein) to train models that balance predictive performance and fairness, and allow for customizable trade-offs. The effectiveness of both the fairness metrics and the composite loss functions is validated through a controlled experimental setup.

Problem

• Addressing group fairness in predictive process monitoring
• Ensuring predictions are independent of sensitive attributes
• Balancing predictive performance and fairness with composite loss

Innovation

• Ensuring predictions unaffected by sensitive attributes
• Using demographic parity metrics like ΔDP
• Composite loss balancing performance and fairness