CS 57100: Artificial Intelligence

Final Project

A significant portion of the course will be a self-directed final team project. The goal is to provide you with an opportunity to get hands-on experience in Artificial Intelligence, practice the techniques and algorithms learned in the course on a problem and environment of interest to you, and even to the extent possible to work on an open research problem. Students should work on the final project in teams of 4 people (we may consider a few teams of 3 or 5, but if we have too many three-person teams we would not have time for final project presentations.) The number of participants in a project, and team organization, will be considered in the evaluation of the project.

Project Topics

Projects are relatively open - they must be centered around some aspect of artificial intelligence that is covered in the course, or that builds on things from the course. The specific problem / topic / technologies to be used are up to you. You may pick any topic, including something related to (or furthering) your research of that of team members - but it must require doing something new, beyond what you might be doing in existing research. Within a topic, some of the things you could do are:

• Application of Existing Artificial Intelligence methods. Identify a new problem (that hasn't been heavily studied in AI) and model it in a way that is amenable to a technique or techniques discussed in class. For example, you may model a problem as satisfiability and use a constraint satisfaction program, or model the uncertainty and use probabilistic reasoning. It may also be interesting to compare two different approaches to solving a problem.
• Develop a new, improved method. Take a problem, perhaps one already well-studied, and develop a new method to address it. (Note that the more well-studied a problem, the harder to identify something new.) You'll need to compare your approach with state-of-the-art.
• Combine multiple AI methods to address a problem that is not amenable to a single method. This may be an extension of existing, reasonably well understood problems - but needs to introduce novel challenges such that existing solutions are insufficient.

This is not an exhaustive list, but meant to give some ideas. You should build on topics in the course - while not fully determined, the list at the course webpage is a good start. In particular, please avoid projects that are based on machine learning (except, perhaps, as incidental to the rest of the work) - machine learning is really the subject of other courses. Projects where machine learning is the core technology/approach are unlikely to be approved.

We suggest you consider a problem where cooperating agents makes sense, as this is an area of AI that you are not as likely to explore in other classes. Using heterogenous cooperating agents also makes it easy for different team members to get a complete vertical view on AI system development.

Part 0: Team. Due September 20

Please submit a list of team members in Brightspace by 11:59pm September 20. If you have three or fewer people, please provide at least a 2-3 sentence sketch of what interests you, as we may assign team members (by September 26) to get to four person teams. If you cannot find team members, or only have two, it is critical to submit a good description of what would interest you so we can assign you to an appropriate team.

Part 1: Proposal. Due October 4

The first part is simply to form your team, identify a general problem you want to address, and a sketch of your initial ideas on how to address the problem. This should be a 2-4 page proposal (typeset), covering at least.

1. An overview of the problem you want to solve.
2. Why the problem is interesting - who cares about it? How is it being solved today? Why could the solution be better?
3. Resources (data, computing, equipment) you will need and where you propose to get it.
4. Risk management: What are the most likely reasons you will be unable to make progress on the project? What will you do if faced with one of these issues?
5. Plan of activities and timeline.

Part 2: Formal Problem definition, background, and approach. Due November 1

The second part is to formally model the problem you will address, and to identify a proposed approach. This includes performing (and describing what you have done) for:

1. Literature survey: Identify and discuss at least three research papers that address the same or a similar problem. Briefly summarize what they have done and how what you are doing is different.
2. Formal model of the problem.
3. Initial solution direction. Describe the methods you plan to use to address the problem, and show how this relates to the formal model.
4. Evaluation plan. How will you determine if your method successfully addresses the problem? What are the measures of success? What are the baselines you expect to beat and aspire to as optimal? Where there are existing AI solutions to the problem, your comparison should include at least one state-of-the-art approach.

Part 3: Final Report and Presentation. Due December 4

The third part requires that you build and evaluate the AI system you propose, and present (in writing and in front of the class) how this addresses the general problem. This involves:

1. Discussion and formal model of the problem (as updated from Part 2.)
2. How you have solved the task (algorithms used, or describe new ones developed) and discussion of outcomes
3. Formally analyze outcomes, including both where they succeed, and limitations. How well do your methods perform relative to baselines / state-of-the-art?
4. Do you have insights that would lead you to a different approach to the problem, or on the methods used that go beyond the problem you have addressed?
5. Discuss the contribution of each team member to the project. If the team does not come to agreement on this portion of the writeup, you may submit an individual addendum separate from the team document.

Note that due to the team nature of the project, and university Quiet Period rules, these deadlines are firm. You may not use late days for any of the project parts.

Each part should be submitted through Gradescope (as a team) as a PDF. Part 1 is maximum 4 pages, Part 2 is maximum 5 pages, and Part 3 is maximum 10 pages. While not required, we encourage you to follow the Springer LNCS format; this will ensure you have sufficient space given the page limits.

For Part 3, you will also need to prepare and submit (as a PDF in Gradescope) a four slide presentation, capturing (in 7 minutes) the problem, key ideas, outcomes, and challenges. You will make a final presentation in class during the last week. While not strictly required, we encourage you to share the presentation responsibilities among all team members (while not always easy, this works very well and gives a great impression when you manage to do it well.)

More details will be provided as the term progresses.