Description:
Data-Driven, Photorealistic Social Face-Trait Encoding, Prediction, and Manipulation Using Deep Neural Networks
Princeton Docket # 19-3597
When one looks at a face, one cannot help but ‘read’ it: in the blink of an eye, people form reliable impressions of both transient psychological states (e.g., happiness) and stable character traits (e.g., trustworthiness). Such impressions are irresistible, formed with high level of consensus, and important for social decisions.
Researchers at Princeton University and Stevens Institute of Technology have together developed a large-scale data-driven methodology that allows for the easy manipulation of social trait information in hyper-realistic face images. This is the very first software design to allow for automatic, photorealistic manipulation of real face images along psychological traits. For example, a given face image could be made to look more or less trustworthy by moving a simple slider. Further, besides generating faces, this method can also ‘read’ faces, providing confidence estimates of perceptions of different social traits for any arbitrary image. And these confidence estimates can provide direct comparisons between images for their perceived psychological qualities including trustworthiness and attractiveness among others. This newly-developed method is both fast and accurate, and represents a paradigm shift in facial photo manipulation.
Applications
- Photo-database search depending on perceived social traits
- Image compression prioritizing specific social traits
- Training away “the other race effect
Advantages
- Low noise and generalizability due to massive datasets
- Manipulation of overall impression of a face image rather than control over individual pixels
- Automatic, photorealistic, and statistically valid manipulation of face images
Intellectual Property & Development Status
Patent protection is pending. Princeton is currently seeking commercial partners for the further development and commercialization of this opportunity
The Inventors
Alexander Todorov is a professor of Psychology in the Department of Psychology at Princeton University. His research interests focus on how people evaluate their environments and how these evaluations shape perception, decisions, and social interactions. He received a Ph.D. in Psychology from New York University, M.S. from New School for Social Research and a B.A. from Sofia University, Bulgaria. He is a fellow of the John Simon Guggenheim Memorial Foundation and Association for Psychological Science, and recipient of the Career Trajectory Award from the Society of Experimental Social Psychology.
Stefan Uddenberg is a postdoctoral fellow working with Professor Todorov at Princeton University. He received his Ph.D. in cognitive psychology from Yale University under the supervision of Professor Brian Scholl and B.A. in cognitive science & Japanese Studies from Dartmouth College. His current research explores the default assumptions wired into the mind, especially in the context of perception.
Joshua Peterson is a postdoctoral fellow in computer science at Princeton University, working with Thomas Griffiths and Alexander Todorov. His work explores the intersection and synergy of cognitive science and machine learning. He received his Ph.D. in cognitive science from University of California, Berkeley and B.A. in Cognitive Science from University of California, Davis.
Thomas Griffiths is the Henry R. Luce Professor of Information Technology, Consciousness, and Culture in the Departments of Psychology and Computer Science at Princeton University. His research interests focus on developing mathematical models of higher level cognition, and understanding the formal principles that underlie our ability to solve the computational problems we face in everyday life. He received his Ph.D. in Psychology, M.A. in Psychology and M.S. in Statistics from Stanford University, and B.A. in Psychology from University of Western Australia. He is a fellow of the John Simon Guggenheim Memorial Foundation and Association for Psychological Science, and recipient of Faculty Early Career Development (CAREER) award.
Jordan Suchow is an assistant professor in the Sch ool of Business at Stevens Institute of Technology. He received his Ph.D. and A.M. in Psychology from Harvard University, and B.S. in Computer Science from Brandeis University.
Contact:
Laurie Tzodikov
Princeton University Office of Technology Licensing
tzodikov@princeton.edu
(609) 258-7256
Daniel Sanchez
Princeton University Office of Technology Licensing
University Administrative Fellow
danielrs@princeton.edu