Building a privacy-preserving federated recommender system for mobile devices

Lerna AI helps app developers to better understand their users and optimize their campaigns. The company’s mobile library optimizes the timing of user engagement, for example by identifying when the right moment is to send an up-selling notification. This is achieved by seamlessly deploying privacy-preserving federated learning on the mobile phone in order to learn from enriched first-party data that never leaves the device.
The project aims at enhancing Lerna AI’s underlying ML algorithms by introducing a recommender system and content-based optimization. This project will primarily assist Lerna AI in improving its predictive capability, i.e. more accurately identify what time and which content are more likely to better engage each individual user. By extension, the project is expected to augment the campaign conversion rates and return on investment for Lerna AI’s own clients.

Faculty Supervisor:

Ioannis Mitliagkas

Student:

Partner:

Lerna

Discipline:

Computer science

Sector:

Information and cultural industries

University:

Université de Montréal

Program: