CSCE 496/896 Project Rules

1. Before you begin your project, you are to submit to sscott@cse and pquint@cse, in text format in the body of an email, a 2–3 paragraph proposal of your project. This is due by Sunday, March 4 at 11:59 pm. You should outline the scope of your project (i.e., what the problem is), what you plan to do for your project, and at least three references (including at least two published conference or journal papers). We will respond soon after with an assessment of your project (i.e., whether or not you may do it) and some suggestions on the scope and possible sources.
2. Check-in reports are due by 11:59 pm on Sunday, April 8.
3. Projects are due by 11:59 pm on Wednesday, April 25.
4. By the project deadline you are to submit, via handin:
   • A project writeup (in a separate pdf file) of approximately 10–15 pages (more on this below)
   • The slides from your presentation (in pdf or PowerPoint)
   • In a zip file:
   • All source code
5. Your writeup should include the following sections.
   • Problem definition
   • Related work
   • Your approach(es), including:
   • details on your architectures, experiments, etc., to enable reproducibility of your results
   • justification of your design decisions
   • what variables you're controlling (e.g., efficacy of a regularizer), and how you controlled them
   • how you measured performance and what you are reporting (type of loss, etc.)
   • Experimental results and your analysis
   • Conclusions and ideas for future work
   • References
6. A note on "Related work": Your project must include a bibliography of at least 4 papers, and a brief discussion (one page is plenty) of their content and relevance to your project. If you do not know of related papers yet, then you might try browsing relevant sections of the text, and/or look through recent conference proceedings such as the International Conference on Machine Learning, Neural Information Processing Systems, International Conference on Learning Representations, and recent journal articles in the journals Machine Learning and Journal of Machine Learning Research. You may also cite the textbook if it gives a good overview of something relevant to your project, but you can do better by reading the papers the book cites.
7. You should also look at Tips on Presenting Technical Material since you will be heavily graded on oral and written presentation of your results.

Last modified 12 March 2018; please report problems to sscott AT cse.