QuantaVerse adds interpretable Machine Learning to enrich AML risk determinations

 QuantaVerse, which uses AI and machine learning to automate financial crime investigations, has enhanced its Financial Crime Investigation Report (FCIR) so AML cases can be accurately adjudicated more efficiently than ever before. Among other critical information, the QuantaVerse FCIR presents investigators an at-a-glance analysis of transactional relationships, negative news, and money laundering typologies along with risk scores calculated by the QuantaVerse machine learning engine and narratives needed to clear a case or document a suspicious activity report (SAR) submission.

The QuantaVerse FCIR, which meets requirements for properly documenting, explaining, and detailing AML investigations, presents the findings  of the QuantaVerse Financial Crime Platform. The QuantaVerse platform has proven to automate approximately 80% of investigation work. By automating the searching, collating, and analyzing of vast amounts of data and then summarizing those findings in the FCIR, QuantaVerse allows investigators to apply their skills, experience, and precious time to adjudicating more cases.

The following enhancements were recently made to the QuantaVerse FCIR:

• Risk scores presented in the FCIR are derived by the QuantaVerse decision engine, a nonrecurrent neural network that interprets observables (such as line of business, adverse media, jurisdiction, entity type, etc.) and makes accurate determinations on risks. QuantaVerse recently applied Local Interpretable Model-Agnostic Explanations, or “LIME,” to further validate risk scores presented in the FCIR.

• Narratives explaining what risks were found in the case are now joined by a new narrative section that explains what risks were able to be cleared. Each set of narratives include the rationale for those determinations which can be used when explaining why a case was cleared or to document a SAR. For investigators tasked with reviewing flagged TMS alerts, the FCIR narratives have helped reduce the average time spent investigating a case by as many as 40 minutes.

• Financial crime investigators using the QuantaVerse system suggested reordering sections of the FCIR to best match the way they work through an investigation. Risk narratives have been moved up earlier in the FCIR and the report’s high-level summary was relocated to the top of the first page to bring investigators up to speed quickly.

“By automating financial crime research and presenting the critical findings in these reports, QuantaVerse has helped reduce by up to 80 percent the time that investigators spend gathering facts and summarizing their research into SARs,” explained David McLaughlin, CEO and Founder of QuantaVerse. “Our goal is to help investigators quickly access critical data, including risk scores calculated by the QuantaVerse machine learning engine, so AML teams can more effectively and efficiently investigate financial crimes. This puts the talents of experienced investigators to the highest and best use and moves us closer to eliminating the $2 trillion of dirty money that flows through our financial institutions every year.”

QuantaVerse offers customers two types of FCIRs. Its alert-based FCIRs examine cases that have been triggered by TMS alerts while entity-based FCIRs analyze and document the risk associated with each customer (and their counterparties) on a regular basis as prescribed by each financial institution.