Multimodal-based inferencing of behavioral constructs in asynchronous video-based interviews: portability, efficacy, and interpretability

The intern is expected to work on devising novel, research-based solutions to model traits (e.g., naturalness, confidence, engagement, humor, etc.) from video recordings of non-cognitive, performance-based tasks. Specifically, we are working toward an intelligent combination of multimodal (video, speech, text) data to create an enriched model for predicting human traits in human-computer interactions. The intern is also expected to assist in the development of on-device, nonverbal tracking that would reduce turnaround time for task-based performance feedback and scoring and preserve personally identifiable information in the mobile device. The intern would gain valuable experience working alongside ETS researchers in an agile environment to create prototypes that have the potential to benefit millions of learners and students worldwide.

Faculty Supervisor:

Jun Chen

Student:

Partner:

ETS Canada

Discipline:

Engineering

Sector:

Education; Professional, scientific and technical services

University:

McMaster University

Program: