An Auditing Framework for Analyzing Fairness of Spatial-Temporal Federated Learning Applications

Federated learning enables remote devices such as smartphones to train statistical models while ensuring that data remains private and secure. Performing privacy-preserving data analysis becomes increasingly crucial as our model is potentially being trained within heterogeneous and massive networks. While federated learning offers the potential to boost diversity in many existing models through on-device learning and enabling a wider range of users to participate, developing fair federated learning models is a challenging task. Throughout this paper, we propose a fairness auditing system for FL models that rely on spatial-temporal data. Borrowing tenets from mobility literature, we propose a set of metrics to define individual fairness using spatial-temporal data. We also introduce a set of approaches for measuring these metrics in distributed settings, as well as building a framework that can monitor the fairness of FL models dynamically.