Advanced Fraud Detection System Real Time Fraud Detection Of Online Payment Frauds
Fuente:
Zenodo
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Recurso digital |
| Veröffentlicht: |
Zenodo
2026
|
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866901938189107200 |
|---|---|
| author | Deepak KV, Bhuvan MM, Beerappa |
| author_facet | Deepak KV, Bhuvan MM, Beerappa |
| contents | <div> <div>We present an end-to-end design and experimental plan for an Advanced real-time fraud detection system for online payments. The approach combines a “Heterogeneous Temporal Graph Neural Network” (HTGNN) for relational and temporal modelling, streaming model updates and approximate neighbour sampling for low latency inference, federated learning for cross institution collaboration without sharing raw data, and explain ability modules for operational investigation. We describe system architecture, deployment options (including in-network inference to meet microsecond/millisecond SLAs), dataset and baseline choices, evaluation metrics, and expected results. Key contributions include (1) a practical HTGNN-based streaming pipeline for transaction graphs, (2) a privacy-preserving federated training and scoring strategy, and (3) an operations plan to balance detection accuracy, latency, and interpretability.</div> </div> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18222783 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Advanced Fraud Detection System Real Time Fraud Detection Of Online Payment Frauds Deepak KV, Bhuvan MM, Beerappa <div> <div>We present an end-to-end design and experimental plan for an Advanced real-time fraud detection system for online payments. The approach combines a “Heterogeneous Temporal Graph Neural Network” (HTGNN) for relational and temporal modelling, streaming model updates and approximate neighbour sampling for low latency inference, federated learning for cross institution collaboration without sharing raw data, and explain ability modules for operational investigation. We describe system architecture, deployment options (including in-network inference to meet microsecond/millisecond SLAs), dataset and baseline choices, evaluation metrics, and expected results. Key contributions include (1) a practical HTGNN-based streaming pipeline for transaction graphs, (2) a privacy-preserving federated training and scoring strategy, and (3) an operations plan to balance detection accuracy, latency, and interpretability.</div> </div> |
| title | Advanced Fraud Detection System Real Time Fraud Detection Of Online Payment Frauds |
| url | https://doi.org/10.5281/zenodo.18222783 |