Researchers have developed Evi-VN, a novel framework designed to improve graph neural networks (GNNs) for fraud detection. Evi-VN addresses the challenge of distinguishing sophisticated fraudsters from legitimate users by focusing on correcting shared "hard regions" where GNNs commonly make errors. The framework injects evidence from various sources, including structured data, text, images, and audio, specifically targeting these difficult cases through virtual class nodes. This approach aims to enhance existing GNNs without disrupting their original design or reliable predictions, as validated across multiple fraud detection tasks. AI
IMPACT This framework could improve the accuracy of fraud detection systems by better identifying sophisticated deceptive accounts.
RANK_REASON The item is an academic paper detailing a new method for graph neural networks in fraud detection. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- Connected Papers
- DagsHub
- Evi-VN
- Gotit.pub
- graph neural network
- Hugging Face
- IArxiv
- Litmaps
- ScienceCast
- scite Smart Citations
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
