Minority Collective Action for User-Side Fairness

TMLR Paper8951 Authors

Abstract: Machine learning models often preserve biases present in training data, leading to unfair treatment of certain minority groups. Despite an array of existing firm-side bias-mitigation techniques, these methods typically incur utility costs and require organizational buy-in. Recognizing that many models rely on user-contributed data, end-users can induce fairness through the framework of Algorithmic Collective Action. In this setting, a coordinated minority group strategically relabels its own data to enhance fairness without altering the firm's training process. We design practical, model-agnostic, minority-only collective-action methods and validate them on real-world datasets. Our findings show that a subgroup of the minority can substantially reduce unfairness with little impact on overall prediction error.

Submission Type: Regular submission (no more than 12 pages of main content)

Assigned Action Editor: ~changjian_shui1

Submission Number: 8951