Communications of the ACM
Stanford ML Tool Streamlines Student Feedback Process for Computer Science Professors
The researchers wanted to focus on helping students grow, rather than just grading their work as right or wrong, so they crafted the language used by the tool very carefully.
Credit: TechSpot
Stanford University researchers have developed and tested a machine learning (ML) teaching tool designed to assist computer science (CS) professors in gauging feedback from large numbers of students.
The tool was developed for Stanford's Code In Place project, in which 1,000 volunteer teachers taught an introductory CS course to 10,000 students worldwide.
The team scaled up feedback using meta-learning, a technique in which an ML system can learn about numerous problems with relatively small volumes of data.
The researchers realized accuracy at or above human levels on 15,000 student submissions, using data from previous iterations of CS courses.
The tool learned from human feedback on just 10% of the total Code In Place assignments, and reviewed the remainder with 98% student satisfaction.
From Stanford News
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found