Projects

Neuroscience

Video Conference Fatigue Experiments

This project, funded by Cisco Research, focused on understanding video conferencing fatigue through neurophysiological experiments. In the post-pandemic world, hybrid work has become the norm, making online meetings an integral part of daily professional life. The study aimed to analyze how specific features of popular video conferencing platforms (e.g., Zoom, Microsoft Teams, Webex), such as self-view on/off and user characteristics like gender influence mental fatigue. Neurophysiological data, including EEG, Heart Rate Variability (HRV), and Skin Conductance (SC), were collected from 32 participants. This data was used to develop a monitoring system that assesses user fatigue and evaluates how platform features affect users’ ability to restore attention.

EEG Processing using LPC-based Methods

This project introduced a novel time-resolved spectral analysis method Linear Predictive Coding Pole Processing (LPCPP), specifically designed for EEG signal analysis. LPCPP offers high noise tolerance and superior time-frequency resolution, making it particularly effective in capturing instantaneous frequency variations in complex signals. Beyond EEG analysis, the method holds promise for integration into AI systems and diverse applications, including speech processing, mechanical vibration analysis, seismographic monitoring, and image analysis, with strong potential to advance real-world Human-Computer Interaction (HCI) technologies.

AI Trustworthy

Trustworthy AI for Recommender Systems

This project, funded by SFI’s National Challenge Fund (OurTech Challenge), aims to develop a simulation framework to investigate how micro-video recommendation algorithms (e.g., YouTube Shorts, TikTok) influence user behavior across diverse user groups. By modeling these platforms, we will evaluate the impact of their algorithms on user engagement, shifting preferences, and potential risks. The framework prioritizes explainability, fairness, harm reduction, and trust, with the overarching goal of informing the design of transparent, user-aligned recommendation systems that foster long-term well-being and trustworthy AI.