Catch more potential online fraud faster with Amazon Fraud Detector
summary
Amazon Fraud Detector is a fully managed ML service that builds customized models from your historical CSVs (minimum 10,000 rows, ≥400 fraud examples), returns risk scores in a few hundred milliseconds, and hosts models for deployment in minutes. Models learn from 20+ years of Amazon fraud signals, show score examples (974, 897) and a practical threshold (≥905 catches ~50% of fraud with ~2% false positives). Customers deploy detectors with rules and orchestration for event-based detection and report production rollout in as little as 30 minutes. No ML expertise required.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:42Demo
- 3:33Agenda
- 3:57The problem
- 6:42Lessons learned
- 12:31Amazon Fraud Detector
- 16:48Demo Walkthrough
- 17:19Demo Web App
- 25:44Creating the detector
- 28:14Examples
related talks
39mFraud Detection: Using ML to Identify and Manage Fraudulent Activities
24mAccelerate your business with AWS AI Services
13mHarnessing Machine Learning to Combat E-commerce Fraud
01h 06mAmazon Machine Learning to Predict Responses
41mUsing Machine Learning to Fight Fraud
36mContinuous Monitoring of Machine Learning Models
42mBuilding a Robust Fraud Detection System
40mFraud detection without labeled data