Quantify and improve the fairness of Federated Learning models
Instruction: Propose methods to measure and enhance fairness in Federated Learning models across diverse clients.
Context: Candidates must discuss approaches to ensure fairness in model outcomes, addressing potential biases in Federated Learning.
Updated
Official answer available
Read the opening below, then unlock the full answer and practical guidance.
I would quantify fairness by measuring performance and error disparities across meaningful client groups, not just across individual examples...