CS570
Aritificial Intelligence
Spring 2008
Due: Friday May 9th
General: Write a paper investigating an advanced version of one of the major topics
we have
covered in the AI course.
Length: 5 pages, space and a half, 12-point font, reasonable margins, etc.
Topics are due:Wednesday April 23rd. This just needs to be a brief paragraph
describing the advanced version.
References: At least 4, of which 2 should be research publicaitons, the others can
be books, magazine articles, etc.
Content: The paper be a detailed review of a successful, or an unsuccessful, advanced version of
one of the topics we've discussed in class (search, game playing, logical programming, etc.).
The paper should include the follow material:
- An explanation of the basic technique.
- An explaination of the advanced technique that is sufficiently detailed that the reader can
understand how it works and, prefereably, implement it.
- A description of the specific sub-domain (if any) that the advanced technique is designed for.
E.g. an advanced game playing search strategy might be specifically designed for games of chance.
- Whether or not in your opinion the advanced technique successfully meets its goals.
Possible topics and starting points for research:
- D* - A version of A* search used when the map is discovered over time [Anthony Stentz,
"Optimal and Efficient Path Planning for Partially-Known Environments," Proceedings IEEE International
Conference on Robotics and Automation, May 1994].
- Cooperative pathfinding - used when multiple agents are finding paths simultaniously, which may
lead to collisions [Silver, David, "Cooperative Pathfinding," Proceedings of the First Conference on
Artificial Intelligence and Interactive Digital Environments, 2005.
- Minmax with learning evaluation functions - adverserial search where the evaluation function is
learned over time [ MINITEX Subscribe (Full Service) Register (Limited Service, Free) Login
Search: The ACM Digital Library The Guide
Feedback
Michael Buro,
"Improving heurisitic mini-max search by supervised learning", Artificial Intelligence,
Volume 134 , Issue 1-2 (January 2002), Pages: 85-99, 2002 ].
- Large scale logical knowledge bases - scaling up logical agents in an attempt to reach "humanish"
intelligence [Sicilia, M.A. Garcia, E. Sanchez, S. Rodriguez, E.,
"On integrating learning object metadata inside the OpenCyc knowledge base"
IEEE Conference on Advanced Learning Technologies, 2004. Proceedings. IEEE International
2004, pgg: 900- 901][http://www.opencyc.org/]
- Fuzzy logic with learning - fuzzy logic where the boundries of the fuzzy categories are learned
over time.
Note that in reading these I will put an emphasis on your analysis of the problem and techniques. Don't just report
what you read, form your own opinion.