Degree
Doctor of Philosophy (PhD)
Department
Computer Science
Document Type
Dissertation
Abstract
Federated Learning (FL) enables decentralized clients to collaboratively train machine learning models without sharing raw data, making it a promising paradigm for privacy-preserving intelligence across large-scale, heterogeneous systems. However, practical FL environments face significant challenges arising from variations in client resources, participation patterns, client behavior, and data distributions. These challenges often lead to inefficiency, unbalanced contributions, and unfairness, ultimately degrading model performance and discouraging long-term client participation. This dissertation advances the state of FL by developing a unified suite of fairness-aware and efficiency-driven frameworks tailored for heterogeneous environments. We investigate fairness from multiple perspectives, including client selection, contribution weighting, and accuracy distribution, and introduce adaptive mechanisms that enable equitable collaboration despite wide disparities in client capabilities. Our contributions span four major aspects: (1) fair client selection and participation balancing mechanisms that incorporate client features, resource constraints, and temporal participation dynamics; (2) adaptive incentive and contribution control frameworks based on game-theoretic modeling of client–server interactions; (3) behavior-aware and dynamic fairness control under bounded-rational client participation, where fairness emerges as an equilibrium of the learning process; and (4) low-rank, communication-efficient federated fine-tuning for large models, where heterogeneous and decaying rank allocation reduces computational cost while preserving fairness across clients. Extensive experimental evaluations on multiple benchmarks under both IID and non-IID settings demonstrate that the proposed frameworks consistently improve fairness metrics, while maintaining or improving final model accuracy compared to state-of-the-art baselines. Collectively, these contributions provide a principled foundation for building fair and efficient FL systems, supporting robust deployment in future heterogeneous edge and distributed environments.
Date
2-6-2026
Recommended Citation
Javaherian, Simin, "Fairness-Aware and Efficient Federated Learning Frameworks for Heterogeneous Systems" (2026). Doctoral Dissertations. 54.
https://scholarshub.louisiana.edu/dissertations/54
DOI
https://proquest.com/docview/3347846497
First Committee Chair
Li Chen
First Committee Member
Anthony Maida
Second Committee Member
Beenish Chaudhry