Faculty Mentor: Sang Won Bae (Stevens Institute of Technology)

Description

Student engagement in the class is essential to improve the effectiveness of learning. Due to the COVID-19 pandemic, on-campus courses have been converted to online courses. During the lecture, the zoom video is usually turned off, and there is a lack of clue whether students can follow the class and learn new knowledge. The problem is that teachers have difficulty knowing whether students are actively engaged in the lecture. While teachers can look around whether students need help in the offline class, it is hard to know whether students are stuck when doing tasks (e.g., coding) in an online course.

In this study, we aim to investigate human cognitive, affective and behavioral engagements by adapting passively sensed facial behavior markers in predicting levels of engagement during the programming exercise, which advance existing categories of measurements such as tracking mouse clicks on a page or monitoring absent time without extra devices (e.g., EEG, eye-motion-tracking, or wearables) in an unobtrusive way.

Objectives and Learning Goals

The students who join this project will achieve the following learning and research objectives:

• Learn the basic concepts of passive sensing technologies, sensors, and data using mobile sensing technologies and applications.
• Explore collected sensor datasets and learn how to preprocess and extract sensor features
• Understand basic statistics and learn machine learning pipelines to develop predictive models
• Investigate libraries (e.g., Python matplotlib), applications, and tools to visualize data and models
• Create user scenario and design prototyping dashboard (e.g., workflow and interface) with a human-centered approach
• Conduct a user study to evaluate the dashboard system.