Overview

A major component of the course is your class project. The goal of the class project is to give you experience in defining, refining and developing a interesting research idea-lette, hands-on experience implementing your proposed research and practice evaluating, writing up and explaining your results. In addition, you will have the opportunity to give feedback to your classmates, in a mini-simulation of the process of the conference submission, reviewing and revising process.

Ideally, your project will be of the quality such that you can submit it to a workshop or conference. Two conference deadlines shortly after the end of the semester are NIPS and ICDM. I will help you try to find an appropriate venue if you decide to submit something based on your project.

You may work in groups of from 2-3 people.

Project Ideas

For examples of some projects done by other machine learning classes, see the following lists:

• Projects from Andrew Ng's class
• Projects from Martin Wainwright's class

You may also want to look at recent machine learning and data mining conferences for ideas:

• International Conference on Machine Learning
• Knowledge Discovery and Data Mining Conference
• Neural and Information Processing Conference
• Siam Data Mining Conference

The above list is incomplete, there are many other venues.

Ideally, your topic will be something that you are interested in, :). If it is related to your thesis research, all the better.

Guidelines

Projects can be either applications oriented -- applying machine learning techniques to an interesting or novel domain/tasks -- or algorithmic/methodological oriented -- comparing several models or developing a new algorithm. In either case, you should validate your results empirically.

Timeline

Mar. 7: Project Proposal due (UPDATED)
Apr. 4: Project Progress Report due
May 2: Project Writeup due
May 11: Final Project Writeup due

There will also be some sort of class presentation or poster session.

Project Proposal

Turn in a 1-2 page proposal describing the domain or problem, the hypothesis you plan to test, the experiments you plan to perform and how you will evaluate the results. Clearly list all of the team member on the project proposal.

You will be commenting on 2 - 3 of your classmates proposals; likewise you will receive comments on your proposal.

Some Useful Resources:

• Marie desJardins' slides on experiment design
• Rob Holte's slides on experimental methodologies
• Ray Mooney's guidelines for a machine learning writeup