CS 5100: Foundations of Artificial Intelligence (Spring 2022) Robert Platt
Project
1 Project Description
The final project should be an application of some area of AI to a problem of interest to you.
For example, you might develop a game playing agent for a simple version of poker, black-
jack, or another card game. The game playing agent might be based on adversarial search
or it might use reinforcement learning. Alternatively, you might use constraint satisfaction
methods to develop various puzzle-solving agents. In the past, students have sometimes ex-
plored applications of classification to machine vision problems. It doesn’t matter so much
what application you choose to study. The key requirement is that you think about how
the methods we studied in class (or perhaps other methods that we did not study!) can be
applied to a real-life problem.
Note that some projects are too ambitious, so make sure you schedule the work to make
sure you have something to show at the end.
We encourage students to work in groups of two or three, but you may also work alone.
Projects with more than three students need explicit permission before a project proposal
can be submitted. If you work in a group, each group member must have a substantial
contribution.
2 Timeline and Deliverables
Project proposal: Please submit an approximately two-page document describing the
problem you intend to address, the algorithms you intend to use, and the comparisons you
intend to perform. We will review all project proposals and enter grades and/or comments
into Canvas. If we give you a grade, then your project proposal is done and you may proceed
on that project. If we do not give you a grade, we will enter comments into speedgrader
regarding how we want you to revise the proposal. Please revise the proposal and resubmit.
The proposal must show that you have read background material on your topic and are
qualified to undertake what you propose to do. It should include full references for the
papers and other sources that you have consulted and that will form the foundation for your
work. If you are working in a group, the planned contribution of each student must be listed
explicitly in the proposal. The proposal must have the following sections:
1. Problem Description: What problem are you solving? Describe the problem formally
from a computational perspective. What are the inputs and outputs (exactly)? What
data are you using (exactly)? Why is it interesting?
2. Approach / Algorithms: What algorithms do you plan to use? Why are these algo-
rithms appropriate? How are these algorithms typically used, and how are you using
1
them? Have other people use similar algorithms to solve your problem before?
3. Planned Comparisons: All projects should involve a comparison of some form. It
could be a comparison between two or more algorithms that solve the same problem, a
comparison between different parameterizations of the same algorithm, or a comparison
between the solutions to different versions of the same problem.
Final Project Presentation. Depending upon how quickly we cover our material this
semester, we may have time for project presentations. I cannot say for sure right now if
we will have presentations or not. If we *do* have presentations, then each presentation
will be approximately 2 minutes. Each team will briefly describe its project and results.
This presentation should contain approximately three or four PowerPoint (or similar) slides
presented via Zoom. Project presentations will be graded but the grade will be incorporated
into the grade for the Final Project Report (below).
Final Project Report. Project reports should be written using the AAAI format (for
the AAAI Conference on Artificial Intelligence): http://www.aaai.org/Publications/
Templates/AuthorKit20.zip. Of course, we don’t expect these projects to be submit-
ted to the conference (although you are certainly welcome to!), it is helpful to look at
papers from previous years to get an idea how they are written. AAAI papers (and pa-
pers from several other conferences) for many years can be found at this link: http:
//www.aaai.org/Library/conferences-library.php. You must submit all code used in
your project along with the project report. However, the report should stand on its own –
we will probably not review your code. Your report should be organized as follows:
1. Abstract: A short summary of what problem you are solving, how you solved it and
what the results are.
2. Introduction: A longer description motivating the problem and approach to solving it.
3. Background: Any background information needed to understand the approach (e.g., a
description the search algorithms relevant to your project).
4. Related work: What other methods could be applied to your problem, why didn’t you
use them and how they relate to your method.
5. Approach: What you actually did in formal detail (with algorithms, equations, etc.).
You should describe the algorithms you used and how they were parameterized.
6. Experiments and Results: A description of the comparison you performed, e.g. which
versions of the algorithm were compared against which other versions? You should
present the results of your comparison quantitatively using graphs. Which version of
the algorithm or which parameterizations worked best? Can you hypothesize why you
had that result? Are there any additional experiments you can run to evaluate this
hypothesis?
2
7. Conclusion: A summary of the results and what you learned by trying to complete this
project.
3