Fraud Detection: Using ML to Identify and Manage Fraudulent Activities
summary
Amazon Fraud Detector is a fully managed service that builds and hosts custom fraud models from uploaded S3 data, augments features (geo, network, issuer), and provides low‑latency real‑time endpoints. The service uses prebuilt templates for online signup and transaction patterns, auto-selects algorithms and tuning, and shows model metrics (example AUC 0.96). A demo run reports ~89% detection and ~2% false positives with ~125 ms average latency. Preview available in us-east-1.
AI-written summary · about
chapters (12)
- 0:00Introduction to fraud detection
- 1:25Common fraud types at Amazon
- 4:13Fraud prevention framework pillars
- 6:01Machine learning in fraud systems
- 9:32Supervised vs unsupervised ML models
- 12:22Challenges of ML fraud detection
- 12:43Amazon Fraud Detector overview
- 18:00Demo creating fraud detector
- 24:42Model performance evaluation
- 27:43Deploying the fraud detector
- 30:39Testing the fraud detector API
- 33:48Additional resources and next steps
related talks
31mCatch more potential online fraud faster with Amazon Fraud Detector
13mHarnessing Machine Learning to Combat E-commerce Fraud
45mAnti-Fraud in Reverse and the Use of ML Methods in It
41mUsing Machine Learning to Fight Fraud
24mAccelerate your business with AWS AI Services
16mLeveraging ML to Create a Reliable Anti-Fraud System
26mAI-Powered Fraud Detection in Cloud-Native Digital Payments
49mEnhanced AML fraud detection solutions with Azure Machine Learning