FACE

FACE

Fraud Analysis Cognitive Engine

Being able to tell when a person is lying is a difficult, but important part of everyday life from interpersonal interactions to life changing criminal sentencing. The ability to detect a lie despite vowing to be truthful, is a significant part of fraud detection. Therefore, we created our Fraud Analysis Cognitive Engine, FACE for short.

Face - Pact™

How FACE Works?

FACE uses vision, audio and text for analysis. The system is trained to recognise micro-expressions, verbal and audio patterns associated with lying. Additionally, the system uses Improved Dense Trajectory (IDT) features, which have been widely used for action recognition and proven effective for detection of deception in video. Voice analysis through Mel-Frequency Cepstral Coefficients (MFCC), also provide a significant boost in performance. Lastly, the scores of classifiers are fused, trained on IDT features, high-level micro-expressions and MFCC to maximise performance.

1
Accuracy

FACE is accurate in 92% of micro expressions. The same test was then given to humans, who were less effective by more than 11%.

2
Support

To support FACE Pact™ has live feeds into Pact™ main platform gaining access to national databases and real time information, delivering enriched data for FACE to make smarter decisions. FACE will revolutionise fraud detection.

How Can FACE Be Used?

Through this unique multi–level automation process, our state-of-the-art AI neural network technology allows us to spread its use across many areas requiring enhanced due diligence, e.g. UK banks are now performing applications for accounts, mortgages and loans via video calling. Through simple API integration FACE can give the bank employee analysis on the interviewee to determine deception and respond accordingly, asking further questions to get more detailed assisted intelligence. This will potentially lead to the bank approving more profitable business and declining potentially toxic business.

Book a Demo

[bookly-form hide=”categories”]