Key
<<<

Tuesday 12pm, 24 April 2018

Understanding students from visual output of graphics-based assignments

Lisa Yan

PhD Candidate - Stanford University

Abstract

As CS classrooms get larger, it becomes more difficult to keep track of students’ learning progress. Grading and feedback are often purely focused on final submissions of student work, a decision often made out of practicality. Furthermore, engaging graphics assignments are particularly difficult to grade, because unit tests often do not cover the space of possible assignments. My work focuses on a combination of statistical analysis and machine learning to tackle these two problems by looking at datasets that capture information about student progress, with a focus on graphics-based assignments. In this talk, I will present the PyramidSnapshot dataset, a dataset of over 100,000 images representing 2000 students’ work progress over a single assignment. I show what types of problems can be solved with this dataset, and what problems remain to be solved. Then I introduce some preliminary work on autograding an assignment modeled off of Breakout, a classic animated graphics Atari game, and introduce some of the challenges related to it. Finally I conclude with perspectives on how I believe intermediate work analysis and computer vision techniques can be used in the classroom.

Bio

Lisa Yan is a PhD candidate in Electrical Engineering at Stanford University advised by Nick McKeown and Chris Piech. Her research is at the intersection of machine learning and computer science education. She is particularly interested in leveraging data to grade and identify struggling students in large introductory computer science courses. Her projects range from developing an effective detection scheme for excessive collaboration and connecting it to learning efficacy to applying computer vision techniques in deep learning to grade and visualize graphics-based introductory programming assignments. Lisa holds an MS in Electrical Engineering from Stanford University, and a BS in Electrical Engineering from Stanford University; she is an NSF Graduate Research Fellow.