CS534 Final Project

Overview
The goal of this class project is to provide you with an opportunity to explore an interesting machine learning problem of your choice in the context of a real-world data set. You are encouraged to combine machine learning research ideas with problems in your own research area. Your class project must be about new things you have done this term, you can't use results you have developed prior to this class. Your project will be worth 25% of your final class grade.
Note that, as with any conference, the page limits are strict! Papers over the limit will not be accepted.

What you need to do

Choose your application domain and learning problem within it and form a team to work on this project. (Projects can be done in teams of up to three people.)

Turn in a project proposal by May 4th. This will give you approximately 5 weeks to work on the project. Feel free to submit the proposal early if you have already decided what you want to do for your project. The proposal should provide the title of the project, the full names of all of your team members, and about a 300-500 word description of what you plan to do. Please send your proposal as a normal email (not as an attachment) to the instructor. If you wish to do a project jointly with another project from a different class you are currently taking (with the consent of the other class' instructor), your proposal must clearly say so.

Turn in a one page milestone report on May 23rd. This is approximately half way between the proposal and final report due date. Your milestone report should describe what you have accomplished so far, and briefly what you plan to do for the remaining part. The milestone report should be written as if this is an "early" draft of your final report, and you should be able to reuse a good portion of this report for your final report.
Give a final presentation on your project (date to be determined, most likely the last day of the class). Each team should prepare a short presentation to describe their work to the whole class.
Turn in a final paper (no longer than 8 pages including references, figures and tables in NIPS format), due June 8th midnight. Each team should turn in a single report and please email me your report before the deadline.

Note that all submissions pertinent to the project (including proposal, milestone, and final report) should be submitted through email to the instructor (xfern@eecs.oregonstate.edu). Please include "CS534-12 Project - XXXX" in your email title, where "XXXX" specify whether it is a proposal, milestone or final report.

How are projects evaluated:

Projects will be evaluated based on:

The technical quality of the work. For example does the technical approach make sense? Are the things tried reasonable? Are the experimental evaluations done in a proper and fair manner? Are the proposed algorithms or applications interesting? Do the authors reveal novel insight about the problem and/or algorithms?
Significance. Did the authors choose an interesting or a "real" problem to work on, or only a small "toy" problem? Is this work likely to be useful and/or have impact?
The novelty of the work, and the clarity of the writeup. Did the authors clearly describe their work such that the technical details could be fully evaluated.

It should be noted that the best class projects often come from students working on topics that excite them. So, pick something that you can get excited and passionate about! If you're unsure what would or would not make a good project, feel free to email me or come to my hours to talk about project ideas.

Project Topics
Projects most often can be grouped into one of the two categories below:

Application project. For this type of projects, you need to pick an application that interests you and explore how to best apply machine learning to solve it. This is typically the most common type of projects that we see in this class.
Algorithmic project. This type of projects focus on developing and testing a new algorithm or some novel extensions of an existing algorithm for solving some machine learning problem. This is most appropriate for students who are doing machine learning research and want to combine their research efforts with this class project.

One could also develop a theoretical project that focuses on proving some interesting non-trivial properties of a new or existing learning algorithm. Projects could also be a combination of these different types, involving both interesting application domains and also new algorithms, or theories.
For inspirations about possible class projects, you might want to take a look at some recent machine learning research papers from machine learning and data mining conferences. In particular, the International Conference on Machine Leanring (ICML) and the ACM SIGKDD Conference on Knowledge Discovery and Data mining are good places to look for such papers. The proceedings of KDD can be accessed through the ACM portal (for 2011 proceeding ---http://dl.acm.org/citation.cfm?id=2020408&coll=DL&dl=ACM&CFID=98900497&CFTOKEN=83398856). The proceedings of ICML 2011 can be found at (http://www.icml-2011.org/abstracts.html).

Some Possible Learning Problems

I am in the process of contact some faculties for potential projects that are directly related to ongoing research at OSU. Once I receive such information, they will be posted here.

Computer Vision Recognition Tasks

Optical Character Recognition

Face Recognition Tasks

Scene recognition (e.g. house vs. no house)
Object recognition in aerial/satellite images
See http://elm.eeng.dcu.ie/~oconaire/cv_datasets.html for a list of computer vision benchmark data sets

Audio Recognition Tasks

Speaker identification
Speaker sentiment
Music genre
Bird species recognition by songs (contact the instructor)

Text Classification and Clustering

Spam
Email Folder Predictor
Newgroup document classifier
Sentiment classification
Author classifier (i.e. take latex files from different authors and try to classify according to author)
There are a few text classification datasets on Andrew McCallum's webpage

Bio-informatics

gene clustering/expression profile analysis
gene sequence analysis

Bird cast