Miniprojects: Brief Excursions into Selected Topics in AI |
CSE 415: Introduction to Artificial Intelligence The University of Washington, Seattle, Autumn 2012 |
Turn-in Requirements.
Due Monday, October 15 (changed from Wednesday, October 10) through
Catalyst CollectIt
at 2:00 PM.
Each team should turn in two files: (1) a PDF, text, or Word document file representing your report, and (2) A Powerpoint or PDF file representing your presentation. The team member whose last name comes first in the alphabet is responsible for turning in the files. |
Purposes.
|
Phases. The miniprojects involve the following phases: (1) partnership and topic assignments, (2) dividing up the work, (3) creating a written report, (4) creating a short presentation, (5) in-class presentation and peer-evaluation for a subset of the miniprojects. |
Partner and Topic Assignments.
The instructor will assign you a partner and topic. However, you can provide
input on these. The instructor will use an automated method (you might call it
one form of artificial intelligence) that will take
your inputs into consideration. However, no two teams will be assigned the
same topic. All the topics should be considered good ones, and all possible
partners also considered good ones.
Stating your preferences is optional. If you do not state any preferences for partners and topics, the AI will use the default value of "neutral" when it considers your preference for any particular partner or topic. However, if you would like to state a preference (positive or negative), consult the list of topics below, log into INFACT, and set your preferences for as many partners and topics as you wish. To log into INFACT enter your UWNetID as your login name and your student number as your password. Note that the topics and possible partners are all on the same page, in a seemingly random order (they are actually alphabetical by username). The topics can be identified by their names being in all CAPITAL letters, whereas the people are not. |
Dividing Up the Work. Communicate with your partner by email, social network, GoPost, telephone, or in person, to reach an agreement on how to split up the topic you are researching, perhaps after you both have done some preliminary reading about the topic online. Decide on a schedule for getting the work done, and a means to collaborate. This could involve a sequence of steps such as this: collecting ideas for the report, drafting the report, critiquing the draft, polishing the draft, making the Powerpoint presentation, checking it, and turning in the documents. Then start on the work. |
Written Report.
Your report should be about 2 pages long. It should contain the following elements:
(1) Title, (2) Authors, (3) Abstract (50-80 words summarizing the topic),
(4) Background and Context for the Idea, System, or Literature Item,
(5) Main ideas explained in a straightforward way, (6) Explanation of
what each team member contributed to the miniproject,
(7) References in standard form. Web references should provide an author name,
if available, a title, a full URL, and a date (e.g., "accessed on October 1, 2012.")
Illustrations and short quotations are welcome, but must be credited to their sources. Do not quote long sections (e.g., chunks of more than 25 words). |
Presentation. Prepare about 2 to 4 Powerpoint slides so that, if selected, you could give a 5-minute presentation on your topic. These can be based on material in your written report. Presentations will take place in class on a future date. |
Topics.
The following topics will be assigned. If you wish, you can state a preference
for a topic (as described above using INFACT).
(Topics in the History of AI) DARTMOUTH CONFERENCE: History of AI: The Dartmough Conference where AI was born MCCARTHYS LISP: History of AI: John McCarthy and the creation of Lisp 1.5 LISP TUTORIALS: History of AI: Early Lisp tutorials: Weissman's Lisp 1.5 Primer / Dan Friedman's TheLittle Lisper ROSENBLATT PERCEPTRONS: History of Neural Nets: Frank Rosenblatt's book Perceptrons MCCULLOGH PITTS History of Neural Nets: the McCullogh and Pitts perceptron model (Topics related to Classic AI systems) SLAGLES SAINT: SAINT symbolic integration program by Slagle BOBROWS STUDENT: STUDENT algebra word problem solver by Bobrow LENATS AM: AM automated mathematical theory developer by Lenat DUDAS PROSPECTOR: PROSPECTOR probabilistic reasoning expert system by Duda et al SHORTLIFFES MYCIN: MYCIN medical expert system by Shortliffe WINOGRADS SHRDLU: SHRDLU microworld plus natural language interface (Topics related to AI languages, platforms and formats) COMMON LISP: Common Lisp -- rationale and major features PRODUCTION SYSTEMS: OPS/5 ordered production system EXPERT SYSTEMS: Use Sacerdoti's survey at http://www.copernican.com/A%20Survey%20of%20Expert%20System LOGIC PROGRAMMING: PROLOG programming in logic -- key ideas (Selected AI Techniques and Topics) CHART PARSING: Chart parsing in natural language processing SCENE LABELING: Scene labeling from line drawings with the Guzman and Waltz methods CONSTRAINT SYSTEMS: Constraint systems and solvers COMMON SENSE: The Cyc project for common-sense knowledge led by Douglas Lenat TURING TESTS: CAPTCHA -- Turing tests for websites SPAM FILTERS: Spam filter design SPEECH UNDERSTANDING: Speech understanding basics |
Updates and Corrections The deadline was changed to Monday, October 15. |