Facial Homophily in Human Social Networks

• Activity Number: 210 - SLDS CSpeed 3
• Type: Contributed
• Date/Time: Tuesday, August 10, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Eric Justin Liu* and Marcus Alexander and Derek Feng and Nicholas A Christakis
• Companies: Yale University and Yale University and Yale University and Yale University
• Keywords: Face Recognition; Social Networks; Homophily; Computer Vision; Friendship

Abstract

In the social sciences, the tendency for people with similar traits to form and maintain social ties is known as homophily. Facial homophily in friendship networks remains poorly studied. We investigate facial homophily within a population of nearly 30,000 from 176 villages in the remote Western Highlands of Honduras, where we mapped approximately 300,000 real-world social connections via complete sociocentric photographic census. Methods for face recognition, landmark detection, and face embedding were used to generate representative feature vectors for each study volunteer. We find that friends tend to have more similar faces compared to strangers. Our research suggests that humans may rely, in part, on facial homophily in organizing into social networks.