Seminar: Personality Sensing: Physiological Signal-Based Personality Detection Using Image and Video Stimuli

Abstract: Personality is an established factor that affects a range of human cognitions and behaviours. However, establishing personality traits typically requires subjects to fill lengthy questionnaires and may occasionally be biased due to the subjects not willing to honestly answer questions. In this work, we address automated personality detection. We propose a novel method that exposes subjects to image and video stimuli and captures their physiological responses using commercial-grade sensors, namely electroencephalogram, galvanic skin response, and blood volume pulse. The responses are processed and fed into a classifier capable of accurately predicting a range of personality traits. We report on the performance of several classifiers and combinations of physiological signals, as well as on the predictive power of image and video stimuli. Our work yields notably high results, raising the potential of robust personality detection for customised human-machine interaction. Bio: Dr Shlomo Berkovsky is the Principal Researcher and Team Leader at the Analytics program of Data61, CSIRO. I study online behaviour of users and their interaction with information, systems, and other users. As such, my work involves aspects of data science, human-computer interaction, and behavioural sciences. My specific research areas include user modelling, personalisation, and persuasive technologies.