GBG Predator with Machine Learning Simplifies and Improves Fraud Detection for Credit Card, Mobile, Digital Payments and Digital Banking Transactions

0

SINGAPORE – Media OutReach – 27
May 2020 – GBG (AIM:GBG), the global technology specialist in fraud and
compliance management, identity verification and location data intelligence,
today announced its expansion of AI and machine learning capabilities for
its transaction and payment monitoring solution, Predator, making deep learning and predictive
analytics available to their entire digital risk management customer journey.
GBG first announced its machine learning capabilities for Instinct Hub, their digital onboarding fraud management system in
January this year. The new AI capability additionally processes third party
data — device fingerprinting, geolocation, mobile and IP, endpoint threat
intelligence, behavioral analytics — assimilated into the GBG Digital Risk
Management and Intelligence platform to enhance their model performance in
fraud detection.

 

With the current pandemic giving rise to changes in
consumer behavior in spending, fund transfers and loans, the ability to
re-learn new data and adapt to new environments can help financial
organizations detect emerging and escalating transaction and payment fraud
trends and mitigate fraud loss. Based on GBG’s “Understanding COVID-19 Fraud
Risks” poll results in April, 37% of respondents see transaction fraud as the
fraud typology that they are most vulnerable to.

 

“Fraud is irregular, complex and evolves dynamically.
Standard fraud model deteriorates over time, exposing businesses to new fraud
typologies and fraud losses. Through continual and autonomous model training in
GBG Machine Learning, we address the issue of model deterioration,” said June
Lee, Managing Director, APAC, GBG.

 

“Today
machine learning provides an average of 20% uplift in fraud detection, GBG
Machine Learning has performed well to provide incremental alerts on missed
frauds for our customers,” adds Lee.

 

GBG Machine Learning utilizes Random Forest, Gradient
Boosting Machine and Neural Networks — three leading and proven algorithms for
fraud detection. These algorithms embody strong predictive analytics, fast
training models and high scalability, learning through both historical and new
data. GBG AutoML  (Automated Machine
Learning) enables adaptive learning to provide the model capability to re-learn
and update itself automatically based on a specified time interval.

 

“Through our APAC COVID-19 fraud risk poll results,
digital retail banking services are growing in demand, from e-wallet, e-loan,
digital onboarding, to digital credit card application; most respondents see a
rise in e-banking services utilization. The ability to easily spot complex
fraud and misused identities in first party bust outs and mule payments, high
volume and high velocity frauds such as online banking account takeover and
card not present frauds across both onboarding and ongoing customer payments
becomes more pressing today,” said Michelle Weatherhead, Operations Director,
APAC, GBG.

 “In addition,
segments like SME lending and microfinancing would be able to harness machine
learning to spot irregularity in borrower patterns by assimilating both
identity, profile and behavioural type data. GBG Machine Learning is able to
analyse large sums of data using algorithmic calculations on multiple features
to determine fraud probability in greater accuracy,” quips Dr Alex Low, Data
Scientist, GBG.

 

GBG Machine Learning is designed to simplify machine
learning deployment for both fraud managers and data scientists, removing the
need to have a data scientist in-house or having to work back to back with the
vendor to lower cost of operation. The solution offers high user controls from
feature creations, model selection and configuration, results and analysis
interpretation and alert thresholds. Users can also configure the solution to
auto schedule and update new fraud patterns through its intuitive user
interface with tool tips built in.

 

The solution takes a “white box” approach to provide an
open and transparent modelling process for ease in model governance and meeting
regulatory requirements. The machine learning score and top contributing
features to results are visible to the users who need to gather further
insights and understanding on the machine learning model performance and
behaviours.

 

About GBG:

GBG offers a series of solutions that help
organizations quickly validate and verify the identity and location of their
customers.

Our innovative technology leads the world in
location intelligence and fraud detection. Our products are built on an
unparalleled breadth of data obtained from over 200 global partners which helps
us to verify the identity of 4.4 billion people globally.

Our headquarters are in the UK and we have
over 1,000 team members across 16 countries. We work with clients in 72
countries including some of the best-known businesses around the world, ranging
from US e-commerce giants to Asia’s biggest banks and European household
brands.

To find out more about how we
help our clients establish trust with their customers, visit www.gbgplc.com/apac, follow us on Twitter @gbgplc or LinkedIn.