Larus Business Automation

Reducing fraud with AI-enabled databases

Graph databases are an alternative to relational databases, which process connections to better understand data. By integrating Fujitsu’s AI technology, LARUS wanted to make its Galileo.XAI solution smarter and faster. Galileo.XAI can also unlock new insights thanks to the Deep Tensor core engine.

Challenge

LARUS wanted to introduce an AI component to its graph-based database platform to enable it to detect new fraud patterns faster with fewer false positives. It turned to Fujitsu for advice.

Solution

LARUS and Fujitsu integrated Finplex EnsemBiz Deep Tensor machine learning technology with Galileo.XAI to create an explainable AI-enabled graph data platform.

Outcomes

  • Fraud detection improved from 72% to 89%
  • False positives reduced by 63%
  • Delivers the speed, scalability, and accuracy required
Fujitsu has really helped drive this project and push the boundaries of innovation. The results have exceeded our expectations.

Lorenzo Speranzoni, CEO, LARUS

72% - 89%

improvement in fraud detection rates

About the customer

Headquartered in Venice, Italy, LARUS helps companies around the world create large-scale, data-driven platforms, based on the latest technologies. As a worldwide Neo4j partner, its certified professionals have been helping universities and companies succeed with the world’s leading graph database for over a decade. LARUS has extensive experience in dif ferent domains, from government, insurance, and financial institutions to manufacturing, retail, and telecom.

Transforming our approach to databases

LARUS specializes in the ecosystem of graph databases, an alternative to traditional relational databases, which store and process connections and relationships to better understand rich data. However, creating a graph-based platform is a complex task because of the algorithms involved.

LARUS developed its Galileo.XAI solution—a graph-based platform for explainable artificial intelligence (AI)—with powerful data analytics and visualization tools. However, the company needed to find the right AI partner to supply the smarts at the heart of the system.

“In the past, we used many dif ferent tools to get results, including notebooks, graph tools, and custom coding,” explains Lorenzo Speranzoni, CEO at LARUS. “Galileo.XAI is the glue which holds everything together to provide an end-to-end, of f-the-shelf graph database solution. And the AI which makes it all work comes from Fujitsu."

Introducing Graph XAI

Fujitsu’s Finplex EnsemBiz offers users functionality ideally suited to financial services applications, with added capabilities based on the Graph XAI technology developed by Fujitsu. Deep Tensor technology can analyze and utilize relationships between graph structure data automatically and ef ficiently, making it possible to detect illegal transactions such as circular trading fraud, which has proven dif ficult with conventional technologies.

Since January 2020, LARUS and Fujitsu have collaborated on a number of proof-of-concept (POC) trials in the financial services space, including POCs aimed at preventing the illegal use of credit cards and illegal automobile insurance claims. This has resulted in increased detection rates and decreased false detection rates.

“Our approach is dif ferent: we don’t see each data point as separate, rather every transaction is seen as part of a larger ecosystem,” adds Speranzoni. “We create context around the data, enabling us to detect patterns that traditional systems miss.”

Boosting fraud detection while reducing false positives

LARUS Galileo.XAI gives its customers the ability to discover common trends that were previously unnoticed, which leads to fresh business insights. It boosts the speed, scalability, and accuracy required by customers such as banks, healthcare organizations, and universities.

As part of the POC, data from credit card companies was used to demonstrate fraud detection in circular transactions. Galileo.XAI has improved fraud detection rates from 72% to 89%, while false positives have reduced by 63%, compared to traditional methods that use manually defined rule bases.

“We can add value to banks’ existing systems and capitalize on their significant fraud investments by identifying new patterns. We eradicate more of the false positives, so we are not wasting time on false alerts,” concludes Speranzoni. “Fujitsu has helped drive this project and push the boundaries of innovation to transform ideas into reality. We have worked together from the outset and the result has exceeded our expectations.”

Related Customer stories

Helsinki University Hospital
Helsinki University Hospital (HUS) has innovated with a partner ecosystem to develop a unique digital solution that makes home dialysis treatment simpler and more sustainable for both patient and professional.
City of Hamm
A co-design workshop with the City of Hamm and Fujitsu identified the most important sustainability data requirements for an open data citizens' portal.
Skogforsk
Enabling Sustainable Forestry for the next 100 years. Using technologies like 5G and VR, Skogforsk has enabled remote control of heavy forest machinery over a private 5G network, enhancing worker safety while boosting productivity and sustainability.
REHAU Industries SE & Co. KG
As the new AI-based Quality Inspection solution eliminates possible defect sources and raises product quality to a new level, it helps to lower costs and make production more sustainable.
HMD Global
The provision of strategic RISE with SAP consultation services removed the burden of having to handle multiple vendors and contracts. HMD Global can now ramp up their manufacturing plants and deliver continuous improvements in business process efficiency, functionality, and automation.
tex.tracer
Blockchain is the basis of the whole tex.tracer structure and fashion supply chain
Township of Barsbüttel
Energy efficiency in public buildings is a top priority for the township of Barsbüttel. To take one step closer to this goal, Fujitsu has captured and evaluated a wide range of room data in real time.