Improving Preference Detection with Eye Movement Gaze and Cognitive Diversity

Preference choices of a subject when given two human face images is a complex cognitive process. One way to detect a subject's preference is by recording and examining the subject's eye movement gaze sequence before his/her decision. Combinatorial fusion analysis / algorithm (CFA) is a new approach for combining multiple scoring systems using rank-score characteristic (RSC) function and cognitive diversity (CD) measure. In this paper, we apply CFA to the study of the eye movement gaze sequences for preference detection. In particular, we use the RSC function to characterize each of the attributes and the CD to measure the diversity between attributes. Our results demonstrate that weighted combination of attributes using diversity strength, computed using average CD's, improves the preference detection.