Artificial Intelligence

G22.2560
Wednesday 5:00-7:00
Room 102, Warren Weaver Hall
Professor Ernest Davis

Reaching Me

Textbook:

Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Norvig
If you're buying the book new, be sure to get the 2nd edition, with the green cover. If you can get the first edition (brown cover) used cheaper, that'll be OK too; I'm not following the book that closely.

Prerequisites:

Fundamental algorithms.

Requirements:

Biweekly problem sets (50%), Final exam (50%).

Description:

There are many cognitive tasks that people can do easily and almost unconsciously but that have proven extremely difficult to program on a computer. Artificial intelligence is the problem of developing computer systems that can carry out these tasks. This course will cover problem solving, automated reasoning, learning, and planning.

There have been two major paradigms for artificial intelligence. Traditional AI has followed the knowledge-based approach, based around handcrafting large symbolic representations of knowledge of the relevant domain. Many of the recent successes of AI, however, have come from the statistical approach, in which simple patterns are automatically extracted from data corpora, and applied to the problem. We will study how both of these paradigms apply to the various AI tasks, the relative strengths and weaknesses of each approach, and possible methods for combining the approaches together.

Course topics:

Instructions for class email list

Link to the class email web page and follow the instructions there for subscribing.

Problem sets

Problem set 1
Solution set 1
Problem set 2
Problem set 3 (PDF)
Solution set 3 (PDF)
Problem set 4 (PDF)
Problem set 5 (Programming assignment 2)

Submitting homework:

Programming assignments must be submitted by email. The format should be the ASCII source file for the code. Be sure to include your name as a comment at the beginning of the code.

Problem sets may be submitted either by email or in hard-copy. Acceptable formats for email are ASCII text, HTML, or Postscript. NO OTHER FORMATS WILL BE ACCEPTED.

Homeworks must be submitted at or before the beginning of class on the day due. Assignments will be accepted up to a week late, with a penalty of one point out of ten. No assignments will be accepted more than a week late.

Class notes

Nondeterministic Algorithms
Propositional Logic
In PDF / In Postscript
First-Order Logic (Revised notes)
Horn clauses
Minimum description length learning

PLEASE NOTE: New message about a typo in problem set 1.

Additional reading (optional)

The Quest for Efficient Boolean Satisfiability Solvers Lintao Zhang, Sharad Malik

Last year's course

You may find the the web site for last year's course useful.

Final Exam

The final exam will be given Wednesday, May 7 from 5:00 to 6:55 in Warren Weaver, room 102.

Sample Final Exam
Solutions to Sample Final Exam
Solutions to Final Exam (Postscript)