Predicting the Occurrence of Déjà Vu using Eye Features
DescriptionIn this work, we investigate the feasibility of automatically identifying when someone is experiencing déjà vu. We collect a dataset of users both experiencing and not experiencing déjà vu, extract eye-gaze features from this sample, and then train a machine learning model to use these eye-gaze features to automatically classify the occurrences of déjà vu. Long term, déjà vu is linked to cognitive processes like familiarity detection, curiosity, and internal memory search – all of which have important implications for educational contexts like intelligent tutoring systems (ITS). We establish that eye-gaze features can be used to automatically identify instances of déjà vu, indicating possibility of identifying these related cognitive states.
TimeThursday, September 14th12:00pm - 1:30pm CDT
LocationTexas Ballroom Foyer