Emotion intelligence plays a significant role in human interactions. But when it comes to human-computer interactions, emotional intelligence still is not widely used. The rapid development of personal digital devices has opened an opportunity to integrate emotion into human-computer interactions. However, there is a gap between existing affective technologies and widespread usage of those technologies. Further, the growing concerns about user data privacy limit the adaptation of such technologies in real life.
This project focuses on bridging the gap between affective technologies and consumer devices while preserving users' data privacy. Primarily, we look into signals that can capture from consumer devices such as smart wearables to identify human affective information without sharing user-generated signals with external services while consuming a small amount of resources on user devices. In technology-wise, we explore representation learning methods and federated learning techniques to bring affect recognition to user devices.

The feature image is using graphics of people & icons form freepick.