Fairness-aware reverse auction-based federated learning
Auction-based Federated Learning (AFL) has garnered significant research attention recently. However, existing methods for AFL data consumers (DCs) primarily focus on improving FL model performance by recruiting DOs with high reputations and low ask prices, disregarding fair treatment for DOs. The c...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Journal Article |
| Language: | English |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/182704 |
| _version_ | 1826112359755677696 |
|---|---|
| author | Tang, Xiaoli Yu, Han |
| author2 | College of Computing and Data Science |
| author_facet | College of Computing and Data Science Tang, Xiaoli Yu, Han |
| author_sort | Tang, Xiaoli |
| collection | NTU |
| description | Auction-based Federated Learning (AFL) has garnered significant research attention recently. However, existing methods for AFL data consumers (DCs) primarily focus on improving FL model performance by recruiting DOs with high reputations and low ask prices, disregarding fair treatment for DOs. The challenge of striking a balance between performance and fairness when recruiting DOs remains unaddressed. To tackle this issue, we propose the Fairness-aware Reverse Auctionbased Federated Learning for DCs (FAR-AFL). FAR-AFL leverages Lyapunov optimization to dynamically adjust selection probabilities for potential DOs, taking into account dynamic changes in participation rates and reputation. FAR-AFL adopts a reverse auction-based DO recruitment mechanism to determine candidate selection and pricing. By combining these components, FAR-AFL improves FL model accuracy while minimizing overall recruitment costs. Crucially, FAR-AFL ensures equitable DO treatment, providing them with fair participation opportunities. Theoretical analysis shows the computational efficiency, individual rationality, and truthfulness of FAR-AFL. Extensive experimental evaluation against six alternative strategies on 4 benchmark datasets demonstrates that FAR-AFL outperforms the best alternative strategy by 1.99%, 6.60%, 1.97% and 23.31% in terms of test accuracy, RMSE, cost reduction and fairness improvement, respectively. |
| first_indexed | 2025-03-09T10:34:16Z |
| format | Journal Article |
| id | ntu-10356/182704 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2025-03-09T10:34:16Z |
| publishDate | 2025 |
| record_format | dspace |
| spelling | ntu-10356/1827042025-02-18T04:18:16Z Fairness-aware reverse auction-based federated learning Tang, Xiaoli Yu, Han College of Computing and Data Science Computer and Information Science Auction-based Federated Learning Lyapunov Optimization Auction-based Federated Learning (AFL) has garnered significant research attention recently. However, existing methods for AFL data consumers (DCs) primarily focus on improving FL model performance by recruiting DOs with high reputations and low ask prices, disregarding fair treatment for DOs. The challenge of striking a balance between performance and fairness when recruiting DOs remains unaddressed. To tackle this issue, we propose the Fairness-aware Reverse Auctionbased Federated Learning for DCs (FAR-AFL). FAR-AFL leverages Lyapunov optimization to dynamically adjust selection probabilities for potential DOs, taking into account dynamic changes in participation rates and reputation. FAR-AFL adopts a reverse auction-based DO recruitment mechanism to determine candidate selection and pricing. By combining these components, FAR-AFL improves FL model accuracy while minimizing overall recruitment costs. Crucially, FAR-AFL ensures equitable DO treatment, providing them with fair participation opportunities. Theoretical analysis shows the computational efficiency, individual rationality, and truthfulness of FAR-AFL. Extensive experimental evaluation against six alternative strategies on 4 benchmark datasets demonstrates that FAR-AFL outperforms the best alternative strategy by 1.99%, 6.60%, 1.97% and 23.31% in terms of test accuracy, RMSE, cost reduction and fairness improvement, respectively. Agency for Science, Technology and Research (A*STAR) Nanyang Technological University National Research Foundation (NRF) This research is supported by the RIE2025 Industry Alignment Fund – Industry Collaboration Projects (IAF-ICP) (Award I2301E0026), administered by A*STAR, as well as supported by Alibaba Group and NTU Singapore through Alibaba-NTU Global e-Sustainability CorpLab (ANGEL); and National Research Foundation, Singapore and DSO National Laboratories under the AI Singapore Programme (AISG Award No: AISG2-RP-2020-019). 2025-02-18T04:18:15Z 2025-02-18T04:18:15Z 2024 Journal Article Tang, X. & Yu, H. (2024). Fairness-aware reverse auction-based federated learning. IEEE Internet of Things Journal, 3504548-. https://dx.doi.org/10.1109/JIOT.2024.3504548 2327-4662 https://hdl.handle.net/10356/182704 10.1109/JIOT.2024.3504548 2-s2.0-85212777934 3504548 en I2301E0026 AISG2-RP-2020-019 IEEE Internet of Things Journal © 2024 IEEE. All rights reserved. |
| spellingShingle | Computer and Information Science Auction-based Federated Learning Lyapunov Optimization Tang, Xiaoli Yu, Han Fairness-aware reverse auction-based federated learning |
| title | Fairness-aware reverse auction-based federated learning |
| title_full | Fairness-aware reverse auction-based federated learning |
| title_fullStr | Fairness-aware reverse auction-based federated learning |
| title_full_unstemmed | Fairness-aware reverse auction-based federated learning |
| title_short | Fairness-aware reverse auction-based federated learning |
| title_sort | fairness aware reverse auction based federated learning |
| topic | Computer and Information Science Auction-based Federated Learning Lyapunov Optimization |
| url | https://hdl.handle.net/10356/182704 |
| work_keys_str_mv | AT tangxiaoli fairnessawarereverseauctionbasedfederatedlearning AT yuhan fairnessawarereverseauctionbasedfederatedlearning |