Samuel Ewusi Dadzie, Franklin College of Arts and Sciences, Statistics, 2nd year PhD student
Statistics Study Assistant uses generative AI technologies to help students learn introductory statistics. The platform features two main components: a question-and-answer (Q&A) system that generates LaTeX-formatted responses to statistics questions, and an image processing tool that converts handwritten or printed mathematical content into editable LaTeX. This integration makes it possible for students to take pictures of their handwritten notes or textbook pages, convert them into LaTeX, and then ask questions about the resulting text.
At the core of the Q&A system is a Retrieval-Augmented Generation (RAG) pipeline, built on OpenAI’s GPT-4. The RAG references a carefully curated knowledge base derived from the creator’s (Samuel Ewusi Dadzie) introductory statistics notes (in LaTeX format). When a student submits a query, the system locates the most relevant sections from these notes and generates a detailed, context-specific answer. This answer is presented as a complete LaTeX-formatted PDF document, which can be downloaded.
The image processing component harnesses GPT-4o Mini’s vision capabilities, enabling it to accurately recognize mathematical symbols and formulas within images in PNG, JPG, GIF, or WEBP formats. By converting the extracted mathematical expressions into LaTeX, students can easily digitize their handwritten work and integrate it into the Q&A process. This removes much of the manual labor involved in transcribing equations or retyping problem sets, freeing students to focus on learning the underlying concepts.
Everything comes together in a user-friendly interface built with Gradio. This interface is divided into three main sections:
- A project information tab, where documentation and background details about the system are provided.
- A Q&A tab, where students can pose questions and receive LaTeX-formatted PDF solutions.
- An image-to-LaTeX converter, which transforms uploaded images of equations into LaTeX code.