Using Markov Chain Monte Carlo with People to Classify Facial Affect
Cognitive science aims to understand how people represent the structure of the world around them. Faces are thought to be windows to some of these representations, namely emotions, which are related to facial expressions biologically and culturally. Labeling expressions is a seemingly effortless task for people, but explaining the subtleties is much more complicated. Jay’s study will help develop a method to systematically explore the scope of different categories of affect, and to explore the correlation between subtle facial movements and the perception of emotion. With sophisticated facial animation software and an algorithm from statistical physics, he hopes to learn to categorize different facial affects through a subject-driven variation of the Markov chain Monte Carlo algorithm. The results of this study will be compared to other contemporary visual categorization techniques.
- Major: Cognitive Science/Statistics
- Mentor: Professor Tom Griffiths, Psychology