Fairness-Aware Federated Learning with Trajectory Shapley Value
2026-05-28 • Machine Learning
Machine Learning
AI summaryⓘ
The authors focus on federated learning, a process where many clients work together to train a model without sharing their private data. They point out that usual methods of combining client updates don't fairly account for how much each client actually helps, especially when clients vary over time. To fix this, they propose a new way to measure each client's real impact on the model's progress, called the Trajectory Shapley Value (TSV). Using TSV, they create FedTSV, a system that adjusts client influence dynamically, which helps the model learn faster, be more reliable, and treat clients more fairly.
Federated LearningModel AggregationShapley ValueOptimization TrajectoryClient ContributionDynamic WeightingPrivacyRobustnessFairnessDistributed Learning
Authors
Daniel Kuznetsov, Ziqi Wang
Abstract
Federated learning is an emerging distributed paradigm that addresses the challenges posed by heterogeneous, privacy-sensitive data. It enables multiple clients to train a model collaboratively by aggregating their local updates at a server. However, conventional aggregation schemes typically use fixed weights that fail to reflect unequal and time-varying client contributions, leading to biased and unstable learning. To improve fairness and stability, we propose the Trajectory Shapley Value (TSV), a contribution metric that evaluates how each client influences the optimization trajectory of the global model using a validation-based, temporally consistent utility. Building on TSV, we design FedTSV, an adaptive aggregation method that converts per-round evaluations into dynamic client weights, allowing the server to respond to heterogeneous and adversarial participation in real time. Experiments on benchmark datasets show that FedTSV accelerates convergence, improves robustness, and yields more equitable contribution assessments, thereby providing a principled foundation for fairness-aware federated optimization.