Privacy + Machine Learning =Peanut Butter + Jelly?

Abstract: Or is it oil + water?

Many question whether it’s possible to enter the machine learning revolution without sacrificing our privacy. In this talk we’ll first explore the conflicts between privacy and machine learning, examining common concerns such as link attacks and profiling. These challenges run deep and have several incompatibilities with the foundations of “big data”. As Paul Ohm famously stated: “Data can be either useful or perfectly anonymous but never both.”

We’ll then shift gears and discuss strategies that allow processing of personal information while protecting privacy. This will include concepts around consent and considering analytical context - a cornerstone of GDPR. Moving into more technical means such as applying differential privacy, privacy by design in all your internal tooling, and visibility into models, such as SR 11-7 guidelines.

By following these strategies, you will be able to adopt machine learning in your organization while meeting regulatory and compliance guidelines, protecting your reputation, and managing the expectations of your data subjects - your customers!

Bio: Steve Touw is co-founder and CTO of Immuta, a unified data management platform that allows organizations to accelerate innovation with confidence. Steve has a long history of designing large-scale geo-temporal analytics across the US Intelligence community, to include some of their very first Hadoop analytics as well as frameworks to manage complex multi-tenant data policy controls. This experience drove he and his co-founders at Immuta to build a software product that frees data science teams to access and work with high-value data. Prior to Immuta, Steve was the CTO of 42six Solutions where he led a large big data services engineering team which was acquired by Computer Sciences Corporation. Prior to 42six Steve led an advanced analytics team for US Special Operations Command, that company, SPADAC, was later acquired by GeoEye and eventually DigitalGlobe. Steve holds a BS in Geography from the University of Maryland.

Open Data Science Conference