Final Project Guidelines

EE511 -- Introduction to Statistical Learning -- Spring 2008

The project is worth a total of 100 points (30% of the grade) allocated as follows:

25 points -- classification performance relative to class mean
20 points -- innovation and soundness of technical approach
25 points -- oral presentation (20 pts based on the average of peer ratings, 5 pts based on Q/A participation)
30 points -- written report (due midnight June 10)

All project materials (results, presentation and report) should all be submitted to the EE511 Projects Dropbox (via UW Catalyst).

Evaluation data will be posted by May 30.
Results are due by noon on June 7.
Results should be submitted to the EE511 Dropbox using the format specified for the project you chose.
The answer key will be posted on June 8, so you can score yourself.

Submit your talk title and a brief abstract (< 60 words) to the EE511 Dropbox, in ascii text or word format, by 5pm June 9.
To keep transition time between speakers short, all presentations will be done on one laptop, so you need to submit your complete presentation to the EE511 Dropbox by 11AM June 10 (in ppt or pdf format).
Presentations will on June 10, 3:30-6:30PM in rm EE 403.
Each speaker will have 5 for their presentation, plus an additional 1-2 minute for questions. Presentations should describe the approach, the reasons behind this choice, and the official results. Contrastive results are highly encouraged.
The professors will give an opening presentation describing the problems and the data, so you do not need to cover that in your individual presentations.
Students in the class are expected to ask at least one question of of one participant.
The students in the class will grade all presentations and optionally provide written comments. We will average these scores for assigning the oral presentation points, and will compile and email any comments (which will be sent anonymously) with our own comments. The comments are intended to help you improve your written report.

A copy of the report should be submitted to the EE511 Dropbox in pdf format by 10AM June 12.
The report should be no more than 10 pages double spaced, including references, figures, tables, etc. This is a maximum - do not feel obligated to make it this long. (Reduce the length by half if you don't double space; use 11pt or 12pt font.)
Use standard bibliographic form when citing other work.
Do not attach source code.
The report should be written in the style of a paper you might submit to a journal and include:
- title and abstract
- introduction (problem statement, motivation, overview)
- background discussion of work you build on or will compare to
- description of algorithm(s)
- description of experimental paradigm (test and training data division, etc)
- experimental results
- discuss results and comment on things that you learned, possibly mention things that would be good to try in the future
- references
The written report will be graded as follows: of the 30 points total, 10 points will be for style (i.e. points will be subtracted for typos, grammatical errors, undefined variables, etc.), 10 points will be for organization and completeness (i.e. addressing all the points listed above), and 10 points will be for clarity in description of the technical approach.

This page is maintained by Mari Ostendorf (mo@ee). Last updated on 2 June 2008.