Human-Machine Intelligent Systems Lab
Lab member

Michael C.

Projects

  • AI in K-12 Math Education Active

    Part of the Gates Foundation-funded MentirAI project, this research builds an AI-powered math tutoring assistant on the historic Mathforum problem-thread archive, exploring RAG (retrieval-augmented generation) and fine-tuning strategies to provide pedagogically sound feedback on student mathematical thinking, in collaboration with Jason Silverman. The work extends into broader investigations of AI applications in educational contexts — including generative AI, classroom integration, and pedagogy — conducted with students across CS394 and the Honors thesis sequence (CS491–CS492). This collaboration began in 2019.

  • NVIZ: Neural Network Visualizer Active

    An interactive, no-code neural network visualizer that lets users upload data, define a network architecture, train a model, and explore its behavior through Sankey diagram visualizations — all in the browser. Developed by Ursinus students Kevin Hoffman and Michael Cummins. Source on GitHub. Related work includes tools for making machine learning model decisions more interpretable and transparent (presented at Ursinus CoSA 2025 by Michael Cummins), and independent studies developing methods to visualize internal feature representations in trained neural networks. Conducted by multiple students across CS391 and CS392. Development began in 2022.

Publications