A Quantum Leap In Fraud Detection Accuracy Using Graph-based Features
summary
High-performance TigerGraph deployments compute graph-derived features (distance-to-fraud, pagerank centrality, louvain community) over terabytes overnight, export them as new columns, and feed them into existing ML pipelines with no model or pipeline changes. Banks running this report a 20% uplift in fraud detection (and claims up to 50%), fewer false positives, and production use at four tier-one US banks; a free try-before-you-buy stack is available at tgcloud.io.
AI-written summary · about
chapters (9)
- 0:00Introduction to fraud detection challenge
- 3:54Limitations of relational databases
- 7:05Graph databases explained
- 11:21Graph analytics for fraud detection
- 17:45Embedding graph features into ML
- 20:10Real-world results and ROI
- 23:23Beyond fraud other use cases
- 25:51How to get started today
- 27:37Q&A and summary
related talks
26mAdvancing Financial Fraud Detection with Graph Dbs
12mMake ML outcomes roar with graph neural networks on TigerGraph and AWS (PRT097)
45mApplications of Graph in Fraud Detection
12mCloud-Native Friendly Fraud Detection
29mTigerGraph, Connecting the World
31mLet's do graph machine learning on Cloud
27mEnhancing Financial Analytics With Event-Driven Architectures and Graph Neural Networks
25mReal-Time Fraud Detection at Scale