Maximizing The Value Of Data With Encrypted Learning

By Ché Wijesinghe, CEO, Cape Privacy

Competitors in any sector are extremely protective of their data for many reasons — among them competitive advantage and brand reputation. The pressing need to safeguard data at all costs can make collaboration between companies impossible. And why would any business want to collaborate with another? The obvious allure is to further enrich first-party data models.

If financial services companies could work with external data sources, such as major retailers, in a secure and trusted way they could derive enormously valuable business insights, increase revenue and drive competitive advantage.

It’s not so far-fetched with the right privacy-preserving technologies.

Marketing Technology News: Why Affiliate Marketers Need Ad Tracking Innovation Now

That’s where encrypted learning comes in.

Financial services firms can build better data models

Financial institutions such as hedge funds and investment management firms rely heavily on machine learning to create the best data models. In fact, 70% of all financial services firms are now using machine learning (Deloitte Insights) and 86% percent of financial services executives plan on increasing their AI-related investments through 2025 (Economist Intelligence Unit).

Using first-party data for modeling is all well and good. However, a 2018 MIT Sloan Management Review data and analytics report found that the most analytically mature organizations use more data sources, including data from customers, vendors, regulators, and competitors. And organizations that share their own data with other firms report increased influence in their business ecosystem, the survey found.

Many financial services firms are willing to pay a premium to subscribe to external data providers to optimize and enrich their own machine learning models. Even a percentage point increase in accuracy of their models can represent millions if not billions of dollars in business value.

Data providers can find new revenue streams

On the other side of the equation, data providers such as major retailers, credit card or payroll companies want to find new revenue streams. Many would be delighted to collaborate with data subscribers in the financial services sector, but of course, these organizations are equally committed to preserving the privacy of their valuable data.

Why encrypted learning matters

Encrypted machine learning changes the narrative on secure and trusted data sharing. It allows data scientists within companies to work with sensitive data without ever decrypting it, thereby protecting privacy by default. With encrypted learning, each organization’s model updates can be encrypted with a unique key and then can be combined and only decrypted in the aggregate.

On the spectrum of privacy-enhancing technologies, encrypted learning has several advantages: it’s more secure than anonymization, it doesn’t affect utility compared to differential privacy, and it is more flexible than federated learning. Performance and scale are clear differences for encrypted learning over other protocols.

Marketing Technology News: No-Code Business Intelligence Tool y42 Raises $2.9 Million to Provide a Scalable & Affordable…

Data science and machine learning has the potential to help solve some of the world’s hardest data problems. Encrypted learning will be considered the new gold standard in privacy-enhancing technology. It is already an absolute game-changer for machine learning —and will have a significant impact on how financial services firms enrich their data to increase the bottom line, all while resting assured privacy is intact.

Beyond Financial Services, encrypted learning has the potential to transcend industries and do for Secure Machine Learning what SSL has done for web browsing.