Predict Responsibly: Improving Fairness and Accuracy by Learning to Defer

Abstract

In many machine learning applications, there are multiple decision-makers involved, both automated and human. The interaction between these agents often goes unaddressed in algorithmic development. In this work, we explore a simple version of this interaction with a two-stage framework containing an automated model and an external decision-maker. The model can choose to say "Pass", and pass the decision downstream, as explored in rejection learning. We extend this concept by proposing "learning to defer", which generalizes rejection learning by considering the effect of other agents in the decision-making process. We propose a learning algorithm which accounts for potential biases held by external decision-makers in a system. Experiments demonstrate that learning to defer can make systems not only more accurate but also less biased. Even when working with inconsistent or biased users, we show that deferring models still greatly improve the accuracy and/or fairness of the entire system.

Accepted as a conference paper at Neural Information Processing Systems 2018

Cited by in corpus (16)

• Fairness in Machine Learning
• Interpretations are useful: penalizing explanations to align neural networks with prior knowledge
• Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty
• Consistent Estimators for Learning to Defer to an Expert
• Review of Mathematical frameworks for Fairness in Machine Learning
• Disaggregated Interventions to Reduce Inequality
• Classification with abstention but without disparities
• Why have a Unified Predictive Uncertainty? Disentangling it using Deep Split Ensembles
• Beyond Individual and Group Fairness
• CheXbreak: Misclassification Identification for Deep Learning Models Interpreting Chest X-rays
• A Bandit Model for Human-Machine Decision Making with Private Information and Opacity
• Preferential Mixture-of-Experts: Interpretable Models that Rely on Human Expertise as much as Possible
• Unifying Model Explainability and Robustness via Machine-Checkable Concepts
• Human-AI Collaboration with Bandit Feedback
• A Sociotechnical View of Algorithmic Fairness
• Style Pooling: Automatic Text Style Obfuscation for Improved Classification Fairness