Developing socially intelligent machines requires adaptation to users' profiles. Research has shown that in the context of robot-mediated learning environments, different learners' profiles can emerge and each of these profiles engage differently with the robot. Previous research has revealed differences in personality traits among different age groups, genders, and even cultures. However, existing methods have focused on one-size-fits-all approaches only and performed social constructs recognition (e.g. engagement and personality) without taking into consideration the user’s profiles.

In this lecture, the speaker presents an approach for personalizing automatic recognition models of personality traits and learners' engagement. The underlying idea is to learn personalized models of self-reported Big Five personality traits, and learners' engagement in two different settings: dyadic Human-Human Interaction (HHI) and educative Human-Robot Interaction (HRI). In particular, the proposed approach automatically learns deep learning architectures for different user profiles using the neural architecture search (NAS) technique for predicting the Big Five personality traits scores and learners' engagement from multimodal behavioral features. 

Speaker

Register now