Fraud Detection with ML/AI: College Technology Center




About the Organization

With a massive system of 100+ colleges, this higher education-focused technology center serves more than 1.5 million students and handles 3.5+ million applications each year. The technology center delivers software solutions and deployments for identity verification, application architecture, networking, and student services to increase college learning outcomes.

The Vision and Challenge

The organization had an issue with fraudulent actors interfering with a reliable application process for prospective and existing students. An email address ending in “edu” allows cybercriminals to reap tax credits, free products and other benefits, as well as achieve further access to students’ sensitive information.

The organization first tried blocking IP addresses and then making changes to the software but neither of these methods was effective and the took a lot of time and effort.

To tackle this issue, the organization sought the help of a trusted third party. They wanted someone with a deep bench of experts specialized in Machine Learning & Artificial Intelligence (ML/AI) to design and implement an efficient process for filtering suspicious activity, which led them to InterVision.

The Outcome

InterVision engaged with the technology center to fully understand the nuances of their IT systems, then helped them leverage predictive analytics and ML/AI to filter their student database system for fraudulent actors by performing analytics and cross-checking on all student information entering its web system using an ML/AI-based engine. As a result, the college systems can immediately identify suspicious activity and flag it for detailed review. This way the IT team can avoid kicking out real students and act against cybercriminals posing as students.

Since the nature of fraud is that it constantly changes, rules-based solutions simply cannot keep up (and require ongoing investment of time from key resources who are already overutilized). One big benefit of ML/AI is that, as the behavior of fraudulent actors changes in favor of new tactics to breach the system, the ML model adapts, learns and continues to thwart their attempts. It’s this ongoing ability to learn and adapt that makes an ML/AI solution so powerful.

The ML/AI solution required high-powered compute resources to re-train the model, so InterVision used extensive automation (Infrastructure as Code) to automate the creation of the AWS ML/AI environment. As a result, high performance environments can be created, and then shutdown once the machine learning model is trained—this approach saved costs for the organization.

Confidence in the application process is a bedrock to a student’s first steps toward education. As a result of the engagement with InterVision, the organization has been able to assure their numerous colleges that all student information is sufficiently protected and verified. The engagement has delivered lowered costs and higher efficiency for the organization compared to a manual process, which means their IT staff can refocus their time on core business projects.

Still a Skeptic of ML/AI?

Download the Skeptic’s Guide to Artificial Intelligence to tell truth from fiction.