STA3000F: Course Project

Overview

The course project is an open-ended research experience in statistical theory or machine learning theory. Its goal is to develop the ability to identify, understand, and investigate meaningful theoretical research questions. The format is inspired in part by an upcoming Mathathon initiative at Caltech.

Projects may take several forms, including:

• a critical synthesis and analysis of existing literature;

• an extension of an existing paper;

• an investigation of an open problem; or

• a new question proposed by the team.

Projects may be completed individually or in teams of two or three students.

Use of Generative AI and Agents

Students may freely use generative AI and agent-based tools to understand the literature, explore ideas, and work toward solutions. The effective use of these tools for theoretical research is an increasingly important skill, and students are encouraged to develop it through the project. Their use is entirely optional: students may also complete the project without them.

Regardless of the tools used, each team is responsible for understanding, checking, and clearly presenting every claim and argument in its final work.

Evaluation

Projects will be evaluated based on:

• the significance of the question and results;

• the completeness and correctness of the work; and

• the quality and clarity of the final report and presentation.

A careful analysis or synthesis of existing literature can earn a satisfactory grade. Meaningful extensions, substantial progress on an open problem, solutions to open problems, or significant new results may receive higher marks according to their importance and level of completion.

Timeline

• Late September: Submit a brief project proposal.

• October: Receive feedback and finalize the topic and research direction.

• December 4: Present the project during the final class.

• End of term: Submit the final report as a PDF. The exact deadline will be announced.