Equifax Uses Deep Neural Machine Learning To Improve Credit Scoring

While artificial intelligence, machine learning and other futuristic-seeming technologies have been resigned to the likes of Apple, Google, Microsoft, Amazon and Facebook, traditional companies are also getting in the game, including Equifax and SAS. According to a report, Equifax is using deep-learning tools to enhance its credit scoring system, and SAS is using deep learning to improve its data mining tools and provide deep learning APIs.

In an interview, Peter Maynard, senior vice president of global analytics at Equifax, said the company realized a few years ago that it wasn’t getting enough “statistical lift” from its traditional credit scoring methods and thus started to embrace advanced deep-learning technology. The report noted that modern machine-learning technologies, such as deep neural networks, which boast much more accurate results, were perceived to not be interpretable, posing a challenge for any company wanting to use them. The complexity also added another layer of challenge for Equifax.

“My team decided to challenge that and find a way to make neural nets interpretable,” said Maynard in the report. “We developed a mathematical proof that shows that we could generate a neural net solution that can be completely interpretable for regulatory purposes. Each of the inputs can map into the hidden layer of the neural network, and we imposed a set of criteria that enable us to interpret the attributes coming into the final model. We stripped apart the black box so we can have an interpretable outcome. That was revolutionary; no one has ever done that before.”

The executive noted that the neural net has improved its ability to make predictive models by as much as 15 percent. The more complex the data, the better the improvement. “In credit scoring, we spend a lot of time creating segments to build a model on. Determining the optimal segment could take sometimes 20 percent of the time that it takes to build a model. In the context of neural nets, those segments are the hidden layers — the neural net does it all for you. The machine is figuring out what are the segments and what are the weights in a segment instead of having an analyst do that. I find it really powerful.”


Latest Insights: 

The Payments 2022 Study: Building A High-Performance Payments Team For Fraud Detection, a PYMNTS collaboration with Stripe, examines how digital platforms of all sectors and sizes plan to develop their anti-fraud teams as part of their their broader growth and development strategies. Drawing from an extensive survey from approximately 250 payments heads at digital platforms in the U.S. and abroad, our study analyzes how poor anti-fraud capabilities can harm platforms’ long-term growth strategies, and how they can build high-performing teams to tackle these challenges.

Click to comment


To Top