Skip to main navigation Skip to search Skip to main content

Cross-Domain Classification of Facial Appearance of Leaders

  • Sungkyunkwan University
  • University of California at Los Angeles

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

People often rely on visual appearance of leaders when evaluating their traits and qualifications. Prior research has demonstrated various effects of thin-slicing inference based on facial appearance in specific events such as elections. By using a machine learning approach, we examine whether the pattern of face-based leadership inference differs in different domains or some facial features are universally preferred across domains. To test the hypothesis, we choose four different domains (business, military, politics, and sports) and analyze facial images of 272 CEOs, 144 4-star generals of U.S. army, 276 U.S. politicians, and 81 head coaches of professional sports teams. By extracting and analyzing facial features, we reveal that facial appearances of leaders are statistically different across the different leadership domains. Based on the identified facial attribute features, we develop a model that can classify the leadership domain, which achieves a high accuracy. The method and model in this paper provide useful resources toward scalable and computational analyses for the studies in social perception.

Original languageEnglish
Title of host publicationSocial Informatics - 12th International Conference, SocInfo 2020, Proceedings
EditorsSamin Aref, Kalina Bontcheva, Marco Braghieri, Frank Dignum, Fosca Giannotti, Francesco Grisolia, Dino Pedreschi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages440-446
Number of pages7
ISBN (Print)9783030609740
DOIs
StatePublished - 2020
Event12th International Conference on Social Informatics, SocInfo 2020 - Pisa, Italy
Duration: 6 Oct 20209 Oct 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12467 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference12th International Conference on Social Informatics, SocInfo 2020
Country/TerritoryItaly
CityPisa
Period6/10/209/10/20

Keywords

  • Face
  • Facial display
  • Leadership
  • Machine learning

Fingerprint

Dive into the research topics of 'Cross-Domain Classification of Facial Appearance of Leaders'. Together they form a unique fingerprint.

Cite this