PROJECT TITLE :

A Siamese Content-Attentive Graph Convolutional Network For Personality Recognition Using Physiology

ABSTRACT:

Affective multimedia information has long been employed as a source of stimulation in the study of a person's personality through physiology. We propose a new Siamese Content-Attentive Graph Convolutional Network (SCA-GCN) to develop a discriminative physiology representation guided by the real video content of emotional stimuli in this paper.

The value of physiology in the task of personality recognition is weighted using the visual content of the stimuli. On a huge public corpus of physiological data, we test our system. In a binary classification for dimensions of Openness, Emotion Stability, and Extraversion, our technique achieves state-of-the-art unweighted accuracy of 72.1 percent, 69.5 percent, and 68.2 percent, respectively, which increases by 20.4 percent, 9 percent, and 13.9 percent over the baseline DNN.

Further investigation demonstrates that the media content has a significant impact on the subject's internal physiological responses, resulting in increased personality recognition performance.