308-557B - Project

Due on the day of the last class of the term by midnight.

The project is meant to be accomplished in teams of two (unless special permission has been obtained). Solo projects are permitted.

Prior to starting your project, it must be proposed in writing by using the project proposal web page (this site will not be available until Feb. 13th) at

New
You should prepare a web page explaining your project (akin to a combination of the proposal, introduction and conclusion in condensed form). This page will be made available to the rest of the class. A template of such a page is available here: http://www.cim.mcgill.ca/~dudek/557/557-project-template.html

The key item that should be submitted and which will be evaluated is a term paper on the project. This must be accompanied by either a videotape, a web page, or some other demonstration medium.

All projects should be based on a paper from the research literature (a journal article) and include an implementation. Exceptions to this format can be discussed in special cases.

The paper should include:

• an introduction;
• a discussion of background and related work including references (and a bibliography),
• a presentation of the method used, its basis, rational, function and assumptions;
• a section on experimental results;
• a conclusion including (important) a detailed critique of the method and an analysis of its strengths and weaknesses.

  Due to scheduling problems in past years, the due dates for the project will have to observed scupulously!

  Evaluation

  Evaluation of the project will be based on several criteria. Essentially, the term paper should resemble a paper from the research literature in form and style. The following issues are among the key features:
  • Good, coherent organization of the paper.
  • A clear well written outline of the method, it's objectives, and the key advantages of the approach.
  • The quality of the implementation, and/or analysis performed as part of the project.
  • The quality of the experimental work, testing, or validation used to evaluate your results, conjectures or code.
  • The critique of the method.

  Caveats

  Every year there are people who do a huge and impressive implementation, but fial to produce an appropriate document. That is a dangerous and tragic strategy. Don't forget to leave time and energy to propery create a good writeup, including citations. Don't forget to seriously consider testing and evaluation of whatever you do! Including only a simple figure from a single test run is an assured way to lose lots of marks.