CS 497 DNR
Computational Theories of Learning and Reasoning

Fall 2001

Course Information

Meeting Times and Locations:

Tue/Thu 11:00-12:15 206 TRANS BLD


Professor: Dan Roth
Office: 2101 DCL
Office Hours: Thursday 1:30-2:30pm
Phone: (217) 244-7068
E-mail: danr@cs.uiuc.edu

Course Description

The purpose of the course is to acquaint students with the theoretical foundations of machine learning and intelligent inference. The focus this year will be on Inference. We will read classical and recent papers that introduce and address important issues in intelligent reasoning. Some of the papers will address issues in knowledge representation and learning, that are important for intelligent inference. The two main foci will be:
  1. The study of knowledge representation and inference formalisms that are amenable to mathematical analysis. In particular, we will look at probabilistic approaches, propositional and relational (FOL) representations, and on different inference formalisms within those. We will also discuss whether, and in what way, different formalisms make sense empirically.
  2. Ways to integrate theories of learning with those of reasoning.
In order to stay with our feet on the ground we will keep in the back of our mind a concrete problem in natural language comprehension, and will try to study how different approaches can be applicable to it.

A more detailed description of topics and a list of papers is given in the Course Plan


The course is targeted at graduates and advanced undergraduates. Ideally, students should have background in basic theory of computation and algorithms, probability, and introductory AI.


The course will not have any exam. Instead there will be several other requirements:
  1. Presenting a paper. Every student will have to present at least one paper. This will require a deep understanding of the paper, relating it to other approaches we discuss and addressing the background problem within the approach presented in the paper.
  2. Writing a short (<1 page) critical survey for each of the papers presented in class.
  3. A term paper. An experimental or theoretical paper in which an inference approach is studied thoroughly with respect to the class' background problem in language comprehension.

Course Materials

Articles will be distributed in class and, if possible, will be available from the course home page http://L2R.cs.uiuc.edu/~danr/Teaching/CS497-01/
Some background material can be found in several books.

Dan Roth