Stevie Chancellor Earns Amazon Grant for Machine Learning and Mental Health Project

Department of Computer Science & Engineering assistant professor Stevie Chancellor is the recipient of the Amazon Research Awards for spring/summer 2022 for her project titled, “Collaborative and socially translucent task instructions for emotionally heavy and subjective annotation tasks.” Her group will use this grant to improve machine learning techniques relating to mental health data. The team is dedicated to creating collaborative labeling interfaces for machine learning modeling while considering the subjectivity of distressing content. 

“One of the challenges of getting good labels for machine learning is the quality of those labels, especially when the task is subjective,” Chancellor said. “This happens when you're trying to label data about people’s mental health. How do you make it so that the labelers themselves don't experience as much emotional weight or graphic content? But also, how do you make it so they can label in a way that helps to deal with subjectivity? Not to get rid of it, but to make it easier for people to label.”

Amazon Research Awards (ARA) provides unrestricted funds and AWS Promotional Credits to academic researchers investigating various research topics in multiple disciplines. Chancellor’s research adds an element of anonymity and collaboration so that labelers are not alone in their work. 

“The goal is to build something that you can know that other people are labeling with you,” Chancellor said. “Normally you don't know that anybody else is doing these annotation tasks. The idea is to make it so you can feel more connected with other people while you're doing this labeling task. By having people be around you while you label, you feel less disconnected and more emotionally supported while you're doing your work.”

The Department of Computer Science & Engineering looks forward to seeing Chancellor’s next steps.