WPI - Computer Science Department, MS Thesis Presentation, Jiani Wang " Deep Multiple-Instance Learning for Stable Attribute Classification in Classroom Video" "

Deep Multiple-Instance Learning for Stable Attribute Classification in Classroom Video

Jiani Wang

MS Student

WPI – Computer Science

Thursday, April 27, 2023

Time: 2:00 p.m.- 3:00 p.m.

Location: Unity Hall 320

Advisor: Prof. Jacob Richard Whitehill

 Reader: Prof. Neil Heffernan

Abstract:

In this paper, we will explore how to classify multiple attributes of each person in a classroom video that are stable over short periods of time, such as their gender, role (student vs.~teacher), and skin tone. This can benefit the field of automatic classroom analysis so as to give teachers better feedback about their teaching and about possible biases they may have towards certain students.

We will tackle this problem using a deep Multiple-Instance Learning (MIL) method. We will also analyze the special aspects of  classroom videos as compared to the kinds of data used in previous MIL research, and also explore strategies for how to deal with it.