ML for Anti-Money Laundering
summary
Claims a payment anomaly detector built in Azure Machine Learning Studio, using one-class SVM and PCA for unsupervised anomaly detection, decision trees for feature logic, and confidential computing to share encrypted datasets across banks. Training initially takes about 12 hours; model is published as a web service for real-time scoring. Reported results are an 85% reduction in false positives and a >200% increase in detected money‑laundering cases, presented as measured outcomes.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:00Money laundering impact
- 8:16Rule-based AML limits
- 10:42AML beyond payments
- 14:37Direct vs indirect AML
- 17:30Dynamic AML systems
- 20:22Data aggregation challenges
- 23:17Feature engineering for AML
- 26:51Data privacy in AML
- 29:47Building AML anomaly detection
- 35:32ML models as cloud services
- 39:31AI AML results
- 40:16Q&A
related talks
49mEnhanced AML fraud detection solutions with Azure Machine Learning
23mChallenges in using machine learning for financial monitoring
18mAnti-Money Laundering ML Modelling Approach
41mLeveraging Automated Machine Learning to enable anyone to develop Machine Learning solutions
32mAI/Machine Learning: What is Actually Working in Cybersecurity?
12mHarnessing AI & LLMs for Next-Gen Digital Payment Security
16mLeveraging ML to Create a Reliable Anti-Fraud System
39mFraud Detection: Using ML to Identify and Manage Fraudulent Activities