This topic involves creating virtual agents (digital human) that can simulate and replicate multi-modal human-style behavioural verbal and non-verbal reactions in response to various external stimulus during human-computer interactions.

This topic creates multi-modal machine learning systyems that can efficiently, objectively, accurately and automatically assess human personality traits and internal mental health status from human long-term behaviours.

This topic create multi-modal machine learning systyems that can efficiently, objectively, accurately and automatically identify image-level human identities (i.e., face recognition), facial expressions, emotions and even multiple facial muscle activations (i.e., Action Units).

This topic focuses on developing various graph representation learning algorithms to convert image, video, audio and other data into graphs with multi-dimensional edge features, as well as various novel GNNs whose nodes can represent matrices or graphs.