Microsoft’s ascent as a software giant is credited in part to the firm’s ability to meet the needs of business customers, from cloud computing to Outlook. Forbes reports that this week, the company has announced a strategic partnership with fintech startup ZestFinance to ease the adoption of AI and machine learning tools for its customers in the financial services sector.
“We’re working with them closely around banking and fraud detection, algorithmic trading, and how they can use AI and deep learning,” said Ed Fandrey, Microsoft’s vice president for financial services, to Forbes.
According to a press release, this collaboration marks “the first fully explainable artificial intelligence (AI) solution for highly regulated industries, starting with the financial sector”. Machine learning models historically have faced obstacles in banking due to their inexplicable “black box” nature, which hinders the necessary transparency in this tightly regulated space.
“The fact is, we haven’t had the ability to use some of the most advanced deep learning models because of the explainability problem. The answer we get with deep learning might be accurate, but in the case of credit applications, that isn’t enough.”
“The fact is, we haven’t had the ability to use some of the most advanced deep learning models because of the explainability problem,” David Berglund, senior vice president of artificial intelligence at U.S. Bank, which is not a Zest customer, told Forbes. “The answer we get with deep learning might be accurate, but in the case of credit applications, that isn’t enough.”
Microsoft is helping Zest solve this explainability problem. According to ZestFinance CEO Douglas Merrill, the company relies on open source technique SHAP (“SHapley Additive exPlanation”). Named in honor of mathematician Lloyd Shapley’s underlying game theory dating back to the 1970’s, SHAP is helping ZestFinance unravel the algorithmic decision-making process and, thereby, facilitate financial underwriting through machine learning.
"Microsoft has built amazing ML services for the enterprise that enable IT organizations to manage and scale their infrastructure easily to meet diverse business needs," says Douglas Merrill, CEO and founder of ZestFinance, in a press release. "With Zest explainability, businesses can start using Microsoft's ML platforms more broadly for high-stakes use cases that demand transparency and auditability."
Merrill believes that machine learning models are the future of credit decisions, enabling financial institutions to provide more credit at lower cost across their entire credit portfolio and serve customers that decades-old legacy models would overlook.
“We want to make fair and transparent credit available to everyone,” he said.