Abadi, Aydin, Naseri, Mohammad, Doyle, Bradley, Gini, Francesco, Guinamard, Kieron, Murakonda, Sasi Kumar, Liddell, Jack, Mellor, Paul, Murdoch, Steven J., Page, Hector, Theodorakopoulos, George  ORCID: https://orcid.org/0000-0003-2701-7809 and Weller, Suzanne
2025.
Starlit: privacy-preserving federated learning to enhance financial fraud detection.
Presented at: 2025 3rd International Conference on Federated Learning Technologies and Applications (FLTA),
Dubrovnik, Croatia,
14-17 October 2025.
2025 3rd International Conference on Federated Learning Technologies and Applications (FLTA).
IEEE,
pp. 126-133.
10.1109/flta67013.2025.11336591
|
Preview |
PDF
- Accepted Post-Print Version
Available under License Creative Commons Attribution. Download (1MB) | Preview |
Abstract
Federated Learning (FL) is a data-minimization approach enabling collaborative model training across diverse clients with local data, avoiding direct data exchange. However, state-of-the-art FL solutions to identify fraudulent financial transactions exhibit a subset of the following limitations. They (1) lack a formal security definition and proof, (2) assume prior freezing of suspicious customers' accounts by financial institutions (limiting the solutions' adoption), (3) scale poorly, involving either O(n2) computationally expensive modular exponentiation (where n is the total number of financial institutions) or highly inefficient fully homomorphic encryption, (4) assume the parties have already completed the entity alignment phase, hence excluding it from the implementation, performance evaluation, and security analysis, and (5) struggle to resist clients' dropouts. This work introduces Starlit, a novel scalable privacy-preserving FL mechanism that overcomes these limitations. It has various applications, such as enhancing financial fraud detection, mitigating terrorism, and enhancing digital health. We implemented Starlit and conducted a thorough performance analysis using synthetic data from a key player in global financial transactions. The evaluation indicates Starlit's scalability, efficiency, and accuracy.
| Item Type: | Conference or Workshop Item - published (Paper) |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Computer Science & Informatics |
| Additional Information: | Rights Retention applied. |
| Publisher: | IEEE |
| ISBN: | 9798331556716 |
| Date of First Compliant Deposit: | 26 January 2026 |
| Last Modified: | 26 Jan 2026 17:00 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/184211 |
Actions (repository staff only)
 |
Edit Item |
Altmetric
Altmetric