Researchers look at protecting businesses from digital fraud
Many businesses are offering their wares through online stores. While that makes for easy purchasing, it also allows more ways for hackers to swindle money. That’s why the Advanced Digital Sciences Center (ADSC), operated by Illinois at Singapore Pte Ltd., is working on a way to detect digital fraud.
“As many more users participate in online activities, the risk exposure to fraud increases,” said ADSC Co-Principal Investigator Carmen Cheh. TFA Geeks—in 2019—reported Southeast Asia lost $260 million (U.S.) to digital fraud. These nefarious activities can be detrimental to small businesses.
Digital fraud comes in many forms. One of them is payment fraud. That is any type of false/illegal transaction done by a cybercriminal. SecurityBrief Asia reported that increased online transactions brought on by the pandemic could result in online payment fraud losses of more than $206 billion between 2021 and 2025.
Another form of digital fraud is promotion abuse. A massive amount of fake accounts is registered “to cash out the promotions,” said Cheh. “Promotion abuse can become even more costly, because anyone with a smart device looking for savings can easily become a promotion fraudster and the numbers can quickly add up.”
Collusion fraud involves fake transactions made within a fraudster organization. Cheh said, “The Association of Certified Fraud Examiners in reporting on collusion in its 2018 Report to the Nations, indicates that there is a direct correlation between collusion and the cost of a fraud incident, rising from an average of $74,000 for one perpetrator to $339,000 for three or more perpetrators.”
The project “Real-Time Deep Learning Networks for Fraud Detection in Modern E-Marketplace Systems”* seeks to combat such frauds. “Our project aims to use state-of-the-art deep learning methods to identify fraudulent transactions and accounts, thus protecting the financial interests of... business owners,” said Cheh.
Their system includes several components. One uses long-short-term memory and graph neural networks to examine the relationships between digital user behavior and transactions. They also use one-class, few-shot learning, and reinforcement learning to look at challenges in digital fraud data samples. Finally, they see how to quickly adapt learning models when it comes to evaluating fraudulent behavior.
Speed is a key factor in their research, said Cheh. “Fraudsters are adapting to fraudulent detection systems at a rapid pace. Thus, we need to build an agile machine learning system that is able to constantly and quickly integrate new knowledge of fraudulent behavior into the detection models. In that way, the detection system will be able to keep up with the evolution of fraudulent behavior in modern e-markets.”
ADSC is collaborating on this project with Professor He Bingsheng (project lead PI) from the National University of Singapore and their industry partner, Chen Jia from Grab, and Professor Chen Deming from the ECE department of University of Illinois Urbana-Champaign.
Illinois at Singapore Pte Ltd. Is a privately held company that is owned by Illinois Global Gateway, an entity of the University of Illinois System.
*This research/project is supported by the National Research Foundation, Singapore under its AI Singapore Programme (AISG Award No: AISG2-TC-2021-002).