10366: Real-Time Fraud Detection with Open-Source AI on IBM z16
Project and Program:
New and Innovative Technologies,
Machine Learning/AI
Tags:
Proceedings,
SHARE Orlando 2024,
2024
In this session, we will show a real customer use case of leveraging popular
open-source and IBM?s proprietary tools to determine which trades and
transactions have a high-risk of exposure before the day?s settlements are made.
We will demonstrate how the AI model developed with the popular AI Open-source
framework TensorFlow leverages data that resides off the IBM Z platform using
Redis. Using Z Container Extension(zCX), this model is deployed on z/OS close to
where the end sample application resides. This setup closely replicates what our
customer has and demonstrates how they achieved high throughput and low latency
by inferencing the AI model on the IBM Z platform -- Presented by Suhas Kashyap
Back to Proceedings File Library