Homework Assignments

There will be 5 homework assignments, approximately one every two weeks (see the class schedule for details on individual assignments). Each assignment consists of building a neural network model for a particular task, using various neural network simulation packages. A major part of the assignment is to evaluate what you did: you will be asked to write a short (one page) report on what you did and how it turned out, how the approach might scale to harder problems, and generally what you learned from the assignment.

The homework assignments are done in groups of two; the idea is NOT to split up the work, but work together so that both of you will learn more. You will turn in one joint report clearly identifying the students who should get credit for it. (In exceptional cases, you may also work alone, but you must justify your case to Risto or the TA ahead of time.)

If you have questions about the assignments, you should go to the TA's office hours or send him email. You will also turn in your work to the TA. Each assignment is due by midnight of the due date, and must be turned in electronically using the turnin program on the departmental unix hosts. As specified in the assignment, top level source code or parameter file listing and execution trace are usually required in addition to the report. Ascii, html, postscript, and pdf are all acceptable formats; Microsoft files (Word, Powerpoint, Excel, rtf, etc) or any other evil proprietary file formats ARE NOT ACCEPTED. If you do not have a unix account in the CS department, you can get one here; setting up the account may take several days so don't wait until the first assignment to request it! If you are unfamiliar with Unix, consider participating in a CS dept unix seminar (they are usually offered during the first couple of weeks of the semester).


risto@cs.utexas.edu
Tue Aug 29 18:12:19 CDT 2006