I was contracted by the FICO product marketing department to review their long term use of artificial intelligence (AI) for fraud, cybersecurity and compliance (AML) solutions. All three of these areas utilize machine learning and AI for anomaly detection. The flagship of this portfolio is the Falcon Fraud Platform, and is used by ~ 10,000 financial institutions to risk score approximately 9,000 payment card transactions/sec globally. I had a blast digging into all that FICO was doing with AI, and I came up with a summary document “5 Keys to Successfully Applying Machine Learning and AI in Enterprise Fraud Detection.” Here is a list of the 5 elements I focused on for the project:
- The role of supervised and unsupervised models in fraud detection (with a focus on behavior anomaly analytics)
- The importance of large data sets in model development and training
- What are predicting features and why is domain expertise necessary in their development
- The benefits of Specialized vs. Generic models in enterprise fraud (i.e., importance of expert features)
- The role of adaptive analytics and/or self-learning AI in enterprise fraud
You can download the report here: 5_Keys_Successfully_Applying_Machine_Learning_ AI
AMULET Analytics 