Steam-powered Turing Machine University of Washington Department of Computer Science & Engineering
 CSE573 Final Mini-project
  CSE Home   About Us    Search    Contact Info 

Administrivia
 Home
 Using course email
 Email archive
 Policies
Content
 Topic Overview
 Slides & Reading
Assignments
 Problem sets
 Project
   

Overview

Instead of a final exam, you should complete a mini-project. It can be on any AI-related topic, including those we have not covered in class. Example topics are listed below.

Collaboration: You can work alone or in groups of two. Groups are expected to do twice the work.

Proposal Date: Send Luke a short email describing your proposed project as soon as you have an idea, but definitely before the end of day on Monday, Nov. 22nd. Please come to office hours, or contact us if you need help deciding on a topic.

Due Date: Friday Dec. 17th, 5pm.

Progress Report: Each group will present a short progress report in class on Thursday, Dec. 9th. The presentations should be approximately 4-5 minutes long, with an additional minute or two for questions. The reports will not be graded, but this is an important opportunity to receive feedback from the course staff and your fellow students. The goal of your presentation should be to briefly describe the problem you are working on, your approach to solving it, progress so far, and your plan for the remaining week. Please come prepared. You are welcome to use the white board. If you would like to use the projector, email pdf slides to Luke on Wed. Dec 8th.

Submit: A final project report and a single compressed file containing source code with instructions describing how it should be run. The project report should be in pdf format with no more than 4 pages of content (you are allowed unlimited space for the cited papers list in the bibliography, starting on page 5). Group projects can have 6 pages of content. You should upload the files to the CSE 573 DropBox.

Project Ideas: A strong project will demonstrate understanding of topics in AI research that are beyond the scope of what we covered in class. This can be done by, for example:

  • Reimplementing an algorithm from a research paper (see list below) and replicating the original results. Although the goal is exact replication, this result can be surprisingly difficult to achieve in practice. A successful project would make significant progress and carefully describe all of the challenges that were encountered.
  • Applying an existing algorithm to a new problem (see list of software below). A strong project would carefully describe the new problem, why the application is appropriate, the results achieved, and include an summary of what was learned from the exercise. Negative results can be interesting if you describe why you originally though the approach would work.
  • Extending the Pacman projects. For example, you might implement and compare POMDP planning algorithms for better action selection in the filtering homework (for an example algorithm, see the RTDP-Bel paper below). You could also implement a range of reinforcement learning algorithms (see the S&B text) and compare performance as a function of problem size / complexity.
  • Other ideas of similar size and complexity are welcome. Feel free to pitch them to Luke if you are unsure.

Research Papers (feel free to suggest others)

Search/MDPs

Games

Beyonds HMMs (other sequence models)

POMDP Planning

E-mail Spam Filtering

Software Packages (feel free to suggest others)

  • SVMlight: Machine Learning with Support Vector Machines
  • Alchemy: A toolkit for Markov Logic Network inference and learning
  • MALLET: Java code, MAchine Learning for LanguagE Toolkit
  • ZMDP: C++ code for scalable MDP and POMDP planning.
  • There is a wide range of Natural Langauge Processing software available from the Stanford and Berkeley NLP groups.


CSE logo Department of Computer Science & Engineering
University of Washington
Box 352350
Seattle, WA  98195-2350
(206) 543-1695 voice, (206) 543-2969 FAX