Preference detection with eye movement

Faculty Mentor: Frank Hsu (Fordham)


Detecting a subject’s (user’s or customer’s) preference over a given set of choices (selections or products) is a common practice in our everyday life and in corporate marketing strategy. This project explores a machine learning method to detect the subject’s preference using eye movement gaze tracking and combinatorial fusion. Two faces, each with its precise coordinates of eyes, nose, mouth, and other regions, are shown to a subject. A sequence of the subject’s eye movement gaze at the two faces, with locations and durations, is recorded and analyzed to detect the subject’s preference using combinatorial fusion, a new paradigm of machine learning, AI, and ensemble method.

Objectives and Learning Goals

After completing this project, participants will have achieved the following learning and research objectives:

  • Learned how to set up an experiment to detect preference of a subject (user or customer) using eye movement gaze tracking devices.
  • Learned how to analyze data set items from the tracked sequence using ML/AI and ensemble methods such as combinatorial fusion.
  • Learned how to generate the attribute set and measure the diversity between attributes.
  • Learned the significant impact of this project on other academic subjects or professional practices such as cognitive neuroscience and marketing informatics.