Federated Learning on Sensitive Mobile Data (Python/Java)

At Lerna IA, we are building a private federated learning platform for mobile applications. Our technology allows mobile-first companies to learn about their app users without violating their privacy. Our novel security architecture speeds up the federated learning process by at least 50x, rendering it practical for real-world mobile set-ups by running the whole ML process on thousands of mobile phones in a distributed fashion. We predict the user behavior based on their context, demographics, mood, etc. in order to identify best timing for out-reach, without retrieving any data!
For the purpose of this internship, the core research portion constitutes the selection and tuning of ML algorithms that are suitable for our application – both useful and with an efficient Federated Learning version.

Faculty Supervisor:

Laurent Charlin

Student:

Partner:

Lerna

Discipline:

Computer science

Sector:

Information and cultural industries

University:

HEC Montréal

Program: