CSE 190 Project Guidelines

CSE 190 Fall 2009 Project guidelines
(cribbed from the excellent guidelines for CSE 254 by Charles Elkan!).

An undergraduate project could be a replication of previous work, but it is preferred if it is not an identical replication! For example, many previous cognitive models have been implemented using back propagation. One relatively do-able project would be to take someone's cognitive model that was previously implemented in backprop and re-implement it with Leabra. Projects that use backprop instead are also allowed, but slightly less interesting! Replicating the results of an innovative recent paper would also be a good, but somewhat more ambitious project. Projects should be closely inspired by one or two specific high quality papers. Novel projects based on your own ideas are allowed, but can often lead to pain and suffering! Here are some ideas for projects.

Projects must involve an implementation of a model. You can use Leabra, or a neural networks package in matlab (see, for example, Jay McClelland's resources page, and his pdptool software). If you do not have access to sufficient computing resources for the projects, contact the instructor immediately.

The schedule for the projects is as follows.

(1) On Thursday, November 12th, you should email me a project proposal (You may email it on Friday, November 13th, but this could be unlucky! ;-). This should explain explicitly and clearly what you will do. In particular, the proposal should include:

your team - single-person projects are possible, but 2-3 team members (e.g., a cognitive science major and a computer scientist) can often make a much better project, if they are well organized.
a list of three to five questions to be answered during the project and discussed in the report
a list of three to five milestones, with a deadline for achieving each milestone
which existing code you will re-use, which software you will modify, and what new code you will write
which datasets you will use and where you can obtain them quickly
the one or two papers that you will build on, with full bibliographic data.

The proposal should be written in well-organized continuous English, as opposed to just an outline. Most of its text should be reusable in your final report. The proposal should be two to four pages long when formatted using either the Cognitive Science Society Word or LaTeX style files. Of course, your paper does not need to be limited to six pages!

(2) Start work immediately on your project. The first phase of the project is especially important. In this phase, some important tasks include:

    * formulating clear, sensible hypotheses to test,
    * finding useful data and designing experiments
    * analyzing and synthesizing closely related recent papers, and
    * selecting and obtaining existing software
    * designing and implementing new software.

These are tasks that can and should be performed mostly in parallel, not sequentially.

(3) On Thursday November 19th, you should hand in a revised and (if necessary) extended project proposal. This must take into account comments received from the instructor and other sources.

(4) On Wednesday, November 25th, you should hand in a progress report. This should be two to three pages long in CogSci format.

(5) On Thursday, December 3rd, you should hand in a draft of your project report. The report should be polished and should resemble a good submission for a cognitive science conference. You should follow the CogSci instructions. It is difficult to find a good set of reviewer criteria for the cognitive science field, but one example of evaluation criteria are the NIPS paper evaluation criteria. There is a section at the end of these on Cognitive Science papers. Read, think carefully about, and follow all the principles of good writing in the "Nuts and Bolts" guide to rhetoric by Michael Harvey. If you don't have time for the whole guide, there is a "nutshell" version.

(5) We will schedule a project presentation time for finals week. There is a final scheduled for the course on Friday (don't worry, there isn't actually a final!) that we can use for the time slot if nothing else works. I would prefer the project presentations to be earlier in the week, but this slot is available: Friday, December 11th, 2009, 11:30AM - 2:30PM

(6) On Thursday December 10th, (finals week), you should hand in the final version of your project report. This will be graded following a revised version of these grading criteria. Perfect academic honesty is required.

The final version of your report should be a Word or PDF file in CogSci format that is no more than a couple of megabytes. If it is much bigger than this, you need to format your figures differently! Word is preferred only because it makes it easy for me to put comments directly in the file using Word's "track changes" feature.

You are expected to spend about eight hours per week on the 190 project, i.e. about 40 hours in total. If you are spending much less time, you are not putting in enough effort and getting a good grade will be difficult. If you are spending much more time, you should think more about efficiency and prioritization.While doing the project, remember that winning at research is similar to winning in many other fields of endeavor.

    * Build on an idea that has been successful in previous work.
    * Make the description of your work understandable, attractive, and memorable; pick catchy names.
    * Keep the work simple. Let the basic ideas shine through.
    * More papers = more likelihood of getting into grad school!