Multimodal Evaluation
Evaluating how AI systems use or fail to use task-critical visual information in curriculum-authentic mathematics.
Worcester Polytechnic Institute
Ethan Croteau
Ph.D. Student in Computer Science
I study image-dependent mathematics: educational math problems where diagrams, graphs, tables, or other visual representations carry information needed to solve the task.
My research is advised by Neil Heffernan and focuses on how AI systems and tutoring infrastructure can represent, evaluate, and respond to visual mathematical information.
Research
Image-Dependent Mathematics
My work examines what happens when mathematical reasoning depends on visual evidence. In these problems, a correct solution may require reading a graph scale, identifying relevant geometric relations, counting visual structure, or connecting labels in a figure to symbolic reasoning.
Alternative Representations
Studying textual, structured, and executable representations of diagrams for AI-supported learning and accessibility.
Tutoring Infrastructure
Exploring problem representations that connect visual facts, mathematical relations, likely errors, and feedback.
Publications
Papers and Outputs
A curated publication list will be added here as my current manuscripts, preprints, and project pages are finalized for public viewing.
View my Google Scholar profileCV
Curriculum Vitae
I am revising my academic CV. A current version will be linked here when it is ready.
Contact
Academic Profiles
- Email ecroteau [at] wpi [dot] edu
- Google Scholar scholar.google.com/citations?user=z24C8vQAAAAJ
- GitHub github.com/ecroteauwpi
- ORCID orcid.org/0000-0002-4142-6353