CSE 574 - Artificial Intelligence II Winter 2004 Modeling Human Intelligence Mon/Wed 1:30 - 2:50 EE1  031 Instructor:  Henry Kautz Reading List & Calendar Email Archive Class List Project Ideas

Image Copyright © 1996-7 by the Department of English, University of Illinois at Urbana-Champaign

Work in Artificial Intelligence has been criticized as being a mere "bag of tricks".  This perception derives from the fact that nearly all publishable papers today concentrate on specific, narrow technical results - a fact is that true of any mature (or near-mature!) scientific discipline.  In this course we will take a step back to look at work in AI from the perspective of the grand challenge of creating computer models of human intelligence.  We will focus on three topics:

• using natural language (not speech recognition or parsing, but rather the pragmatics of language)
• probabilistic models of human behavior
• creativity

Course Work

Students will each lead the discussion of some of the papers we read in the course and create a course project.  The course project may either be a paper (typically 15 pages) treating some topic related to the course in depth, or a programming project that may be done individually or in a group.  In either case the project will be submitted iteratively - proposal, first draft, final draft - according to a timeline to be posted.

Syllabus

• part 1: discourse understanding
  • speech act theory
  • beliefs about beliefs and goal
  • planning utterances
  • interpreting utterances
  • the structure of discourse
  • reinforcement learning in discourse analysis
• part 2: behavior recognition
  • technical foundations: from Markov models to Dynamic Bayes Nets
  • modeling events with structure and continuous time
  • learning user models
  • modeling user errors and emotions
  • applications
• part 3: creativity & emotion
  • theories of creativity
  • computers that create art and music
  • emotional computers