Research in Modeling Emotions in Interactive and Multimodal Contexts

• Lead the development of a large-scale multimodal interaction corpus capturing co-speech gestures, full-body motion, and expressive speech during dyadic conversations, in collaboration with experts in acting and behavioral sciences.
• Design affective computing frameworks that move beyond static, unimodal processing by modeling dynamic emotion co-regulation, turn-taking cues, and gestural-prosodic synchrony in real-time interactions.

Research in Trustworthy AI on Emotion Recognition and Healthcare

• Advance fairness-aware learning across heterogeneous signals. Design and evaluate fairness-aware models for both speech and physiological signals, tackling demographic imbalance and individual variability.
• Mitigate reliability challenges under real-world data constraints and address incomplete or noisy input conditions through adaptive model design.
• Investigate how fairness and uncertainty modeling contribute to the trustworthiness of affective systems deployed in sensitive domains such as healthcare and mental wellbeing.