Affective Computing & Human-AI Interaction

Multimodal Emotion Recognition Through Physiological and Behavioral Integration

This research investigates how artificial systems can accurately detect and classify human emotions by integrating multiple data streams including facial expressions, vocal acoustics, body posture, and physiological signals like heart rate variability and skin conductance. By developing robust fusion algorithms and deep learning architectures, this work advances our understanding of emotion as a complex, multidimensional phenomenon that cannot be captured through single modalities alone.

Neural Mechanisms of Affective AI Alignment With Human Values

This research explores how artificial neural networks can be designed to recognize, understand, and respect human emotional and value systems during interaction. The scientific contribution lies in establishing computational frameworks that bridge affective neuroscience principles with machine learning, enabling AI systems to make decisions aligned with human emotional wellbeing and ethical values.

Real-Time Sentiment Dynamics in Human-AI Conversational Agents

This research examines how sentiment and emotional valence evolve dynamically throughout extended human-AI dialogues and what triggers affective shifts. The scientific insight generated reveals feedback loops between user emotional expression and AI response strategies, leading to computational models of emotional scaffolding and empathetic dialogue design.

Affective State Transfer Learning Across Diverse Cultural Populations

This research investigates how emotion recognition models trained on one cultural population can be adapted and generalized to other cultures without significant performance degradation. The academic contribution addresses fundamental questions about universal versus culture-specific affective expression patterns and develops novel domain adaptation techniques for cross-cultural affective AI.

Adversarial Robustness of Emotion Detection Systems Against Spoofing

This research studies how emotion recognition algorithms are vulnerable to deliberate manipulation and develops defensive mechanisms to ensure affective AI reliability. The scientific discovery reveals specific vulnerabilities in biometric emotion sensing and produces hardened deep learning architectures capable of distinguishing genuine from artificially-induced affective signals.

Computational Models of Empathetic Response Generation in Social AI

This research develops and tests computational frameworks that enable AI systems to generate contextually appropriate, emotionally-resonant responses that demonstrate understanding of user emotional states. The scientific contribution establishes measurable constructs of computational empathy and validates whether machine-generated empathetic responses produce measurable improvements in human trust and wellbeing.

Temporal Dynamics of Affective Contagion in Human-AI Teams

This research examines whether and how emotional states spread between human team members and AI agents during collaborative tasks, and what mechanisms facilitate or inhibit this contagion. The investigation produces temporal models of affective influence in mixed human-AI systems and reveals design principles for managing emotional dynamics in human-centered AI collaboration.

Neurofeedback-Informed AI Systems for Emotion Regulation and Mental Health

This research integrates real-time neuroimaging and physiological monitoring with adaptive AI systems to deliver personalized emotion regulation interventions. The scientific advancement establishes closed-loop affective computing systems grounded in neuroscience principles that can dynamically adjust support based on objective neural markers of emotional state.

Ethical Dimensions of Affective Manipulation Detection in AI Systems

This research develops methods to identify and mitigate intentional or unintentional emotional manipulation by AI systems and establishes ethical frameworks for responsible affective computing. The contribution produces auditable affective AI systems with transparent emotional reasoning and accountability mechanisms that protect users from coercive or exploitative affect-based influence.

Predictive Modeling of Long-Term User Engagement From Affective Signals

This research develops machine learning models that predict sustained human engagement, satisfaction, and relationship quality with AI systems based on early affective interaction patterns. The scientific insight reveals which emotional dynamics during initial human-AI encounters serve as robust predictors of long-term interaction quality and user retention.