Xiaodong Qu Research Group

George Washington University

Can machine learning remain reliable and interpretable across diverse human neural and behavioral data? The group develops machine-learning methods for noisy, temporal EEG and multimodal human data to improve generalization across people and sessions. Researchers build human-in-the-loop and interactive EEG/BCI systems, including real-time personalization and accessible EEG game control, to study model behavior under user adaptation. The lab evaluates temporal and multimodal modeling techniques (convolutional and transformer approaches) for EEG tasks such as gaze prediction and motor imagery. Projects examine AI for learning by measuring EEG markers during learning and testing generative-AI effects on higher-education tasks.