Last week, LARUS Team together with its Partner Fujitsu Labs of America joined Nodes 2021, the Neo4j Online Developer Expo and Summit. An occasion to be part of one of the greatest global Neo4j events during the year.
We had the opportunity to be speakers in the “Best Practice” track, in order to explain how to harness Graph-AI to combat Fraud in Fintech and Insurtech sectors with our joint solution Galileo XAI, launched for the first time right at Nodes 2020 in last November.
Alberto De Lazzari, LARUS Chief Scientist and Surya Josyula, Marketing Director at Fujitsu Labs of America talked about frauds. Especially in their dynamic nature, they are a major area of concern requiring significant time and resources to isolate from an enormous volume of transaction data. Our innovative new composite AI based solution, that combines graph-rule-based with graph-supervised-learning coupled with explainability, addresses this problem.
The speech was based on a real world Graph AI (GAI) project undertaken by LARUS, Inc. and Fujitsu Research. The project evaluated the incorporation of GAI technology from Fujitsu (Deep Tensor) into an existing rule-based insurance fraud detection application developed by LARUS. Since the application was in production, we were able to obtain a sample of 8,700 accident cases that triggered the alert rules and were manually reviewed by insurance experts. Using this sample as labeled data to train supervised machine learning models, we found out that the algorithm used to partition the source graph into labeled subgraphs (“communities”) had the largest impact on the final results. This graph supervised learning is able to adapt better to fraudsters changing behaviour compared to fixed rules.
Moreover, during the event, we were pleased and proud to show the new logo of Galileo XAI, the new restyling about the insight data platform and launched the new website www.galileox.ai: now you can discover more about the platform on it and stay up to date. A big step for the Galileo XAI Team which we’re very happy about.
Nodes 2021 was an incredible day for the global graph community, with talks on innovative and exciting graph projects. Do you want to relive all the awesome sessions or did you miss the event? Watch our talk into Best Practice track and check out the ones you might have missed with our on-demand NODES Library.