Fourth University Machine Learning Research Team Receives MathNet Award

Dotted blue decorative line - ASSISTments Images
ASSISTments will provide four university machine learning research teams with access to the largest dataset of digitized student work for the purpose of designing AI enabled technology to read, analyze, and score digitized student math work, and provide learning insights to math teachers.

A Fourth Machine Learning Research Team Has Been Awarded Access to the Largest Dataset of Digitized Student Work

WORCESTER, MA, March 14, 2023—ASSISTments, a nonprofit organization founded in learning science, recently announced the winners of the first MathNet Competition sponsored by Doug Jaffe and Laurence Holt. In conjunction with the MathNet awards, ASSISTments will provide four university machine learning research teams with access to the largest dataset of digitized student work for the purpose of designing AI enabled technology to read, analyze, and score digitized student math work, and provide learning insights to math teachers.

MathNet Competition grants have been awarded to the following research teams:

  • Kent State University: PI: Karl W. Kosko, Co-PIs: Dr. Qiang Guan, Dr. Richard E. Ferdig
  • University of Florida: PI: Dr. Wanli Xing, Co-PIs: Dr. Anthony Botelho, Dr. Jinnie Shin
  • Worcester Polytechnic Institute: PI: Dr. Neil Heffernan, Co-PIs: Sami Baral, Abhishek Santhanam, Dr. Li Cheng
  • Rutgers University: PI: Dr. Dake Zhang, Co-PIs: Co-PI: Dr. Min Li, Dr. Dong Deng, Advisers: Prof. Robert Siegler, Prof. Jingang Yi (new awardee)

Grantees will have access to the ASSISTments dataset which includes over 3 Million images of student work spanning from grade 1 to high school algebra and geometry.

“We envision a future in which K-12 mathematics education products support educators, students, and families with automated analysis of written student work,” commented Cristina Heffernan, Executive Director of The ASSISTments Foundation. “We are thrilled with the high-caliber machine learning research teams and anticipate results that, when applied to the classroom, will improve real-time diagnostic assessment and feedback to students and teachers completing mathematics tasks.”

“This project keeps the focus on student work completed as part of the math curriculum.  These technologies will help educators focus time on the student papers most in need of attention while ensuring that all of the students' work can be considered when assessing how each student in the class is progressing,” commented Doug Jaffe, MathNet Co-Founder.

About the MathNet Teams and their Projects

1. Kent State University Team

  • The Team: PI: Karl W. Kosko, Co-PIs: Dr. Qiang Guan, Dr. Richard E. Ferdig
  • Project Title: Designing A Machine Learning Tool To Assess Children’s Fraction Arithmetic
  • Goal: This project aims to design and pilot a tool that allows teachers to take a picture and/or scan with their smartphone or tablet to receive an Machine Learning based report assessing the student’s response, procedure, and level of reasoning with fractions.

2. University of Florida Team

  • The Team: PI: Dr. Wanli Xing, Co-PIs: Dr. Anthony Botelho, Dr. Jinnie Shin
  • Project Title: Project ViTAMR: Vision Transformer-based Augmentation through Diagnostic Feedback to Support Mathematical Reasoning
  • Goal: Project ViTAMR /vīt-ə-mər/ aims to generate new insights and understanding on students’ math learning strategies with hand-worked responses (e.g. images and texts) for mathematical reasoning and further create a machine learning pipeline using multimodal and multi task deep learning models to provide automatic diagnostic feedback for students’ mathematical reasoning. Diagnostic feedback has been shown to be a valuable pedagogical strategy to invoke, assist, and train students’ mathematical reasoning.

3.  Worcester Polytechnic Institute Team

  • The Team: PI: Dr. Neil Heffernan, Co-PIs: Sami Baral, Abhishek Santhanam, Dr. Li Cheng
  • Project Title: Leveraging Artificial Intelligence to Analyze Students’ Math Work Uploaded in a Digital Platform
  • Goal: 1) Release a set of images (hopefully 100,000) that have PII stripped from them (students’ faces and their names that they might have written on the piece of paper and inadvertently included in their uploaded image) and publish the first paper using this data to spur the field to "beat" Dr. Heffernan and team; 2) Developing autoscoring techniques to facilitate teachers in assessment of student work with images.

4.  Rutgers University Team

  • The Team: PI: Dr. Dake Zhang, Co-PIs: Co-PI: Dr. Min Li, Dr. Dong Deng, Advisers: Prof. Robert Siegler, Prof. Jingang Yi
  • Project Title: Automated Classification of Student Problem-Solving Style in Representing Fractions with a Number Line
  • Goal: 1) An AI goal: to develop an innovative approach to provide automatic classifications of students’ problem-solving styles (i.e., considering accuracy, strategy choice, error type) in representing fractions with number lines, using machine learning techniques, 2) A Psychometric goal: to model how item features are related to students' performance on number line representation of fractions, and 3) An Educational Psychology goal: to model how item features and student characteristics/prerequisite math skills (e.g., division skills, spatial skills) interactively affect student problem-solving style classifications.

Learn more about the MathNet competition on the E-TRIALS website.

About The ASSISTments Foundation

The ASSISTments Foundation is a 501(c)(3) non-profit organization established in 2019 with generous funding from the Chan-Zuckerberg Initiative and Schmidt Futures, and sponsored by Worcester Polytechnic Institute (WPI). The organization is dedicated to scaling the reach and impact of ASSISTments in classrooms nationwide, driven by the belief that every student deserves the opportunity to be good at math. The ASSISTments Foundation works with WPI to conduct cutting edge research on the learning sciences. www.assistments.org

MathNet

About Doug Jaffe

Doug is a consultant and advisor to philanthropies and non-profits in the K-12 education sector. He was a Senior Fellow at the Bill & Melinda Gates Foundation and held leadership roles at the New York State Regents Research Fund and the New York City Department of Education where he led human capital, strategy, data and technology initiatives. Prior to that Doug held roles in management consulting and product development in the private sector.  Doug earned his J.D. from Columbia Law School and is an alumnus of the Broad Residency in Urban Education.

About Laurence Holt

Laurence works with philanthropic organizations on innovations in K-12 education. Before that he was Chief Product Officer at Amplify where he oversaw the engineering, design and production of the company's curriculum and assessment solutions. Over ten million students use those products today. Prior to that, Laurence founded and ran a business and technology consulting firm in London and New York with clients including Apple Computer, the BBC, and Goldman Sachs. He wrote three award-winning how-to books on topics including recognizing constellations and analyzing stock charts. He lives in Paris, France with his spouse, an artist, and ten-year-old twins.

Share on socials!

Facebook logoLinkedIn LogoPinterest logo
View All Blogs