CS 3600
Intelligent Systems

syllabus3600.pdf

Prof: Mark Riedl
riedl @ cc
TSRB 220
Office Hours: MWF 4:30-5:30

TA:Richard Roberts
richard @ cc
CCB 2nd Floor
Office Hours: T 3-4, W 1:30-2:30

TA: Adity Dokania
adity.d @ gatech
TSRB 225
Office Hours: M 11:30-12:30, F 1:30-2:45

Introduction to Intelligent Systems is a three-credit undergraduate course on Artificial Intelligence. The class is called Introduction to Intelligent Systems instead of Introduction to Artificial Intelligence to emphasize that we intend to approach AI from the point of view of building intelligent agents, environments and systems. In particular, you will learn about the methods and tools that will allow you to build complete systems that can interact intelligently with their environment by learning and reasoning about the world.

There are three primary objectives for the course:

• To provide a broad survey of AI and Intelligent Systems
• To develop a deeper understanding of several major topics in AI
• To develop the design and programming skills that will help you to build intelligent artifacts

In practice, you should develop enough basic skills and background that you can pursue any desire you have to learn more about specific areas in IS, whether those areas are planning, knowledge representation, machine learning, vision, robotics or whatever. In particular, this class provides a useful foundation for a number of courses involving intelligence systems, including Machine Learning (CS4641), Knowledge-Based AI (CS4634), Computer Vision (CS4495), Robotics and Perception (CS4632), Natural Language Understanding (CS4650) and Game AI (CS4731).

Someone once said that the trick to doing AI is coming up with a good representation. That's not quite all there is to it, but it's close enough, so to succeed at this class, you should know a bit about data structures and algorithms. At the very least, you will have to be able to read pseudocode and understand basic algorithms as they are presented to you.

Someone else one defined AI as finding fast algorithms for NP-hard problems. Again, that's not quite all there is to it, but it's not too far from the truth, so it also turns out that a familiarity with (or at least a lack of abject fear over) some basic theory helps to situate many of the algorithms.

As the semester continues, it turns out that a familiarity with basic probability theory will also be very useful; however, we will spend some time on that in class in order to refresh your memory. Finally, you should feel pretty comfortable programming on your own. Many of the projects will be in LISP, and perhaps one or two will be in C. We will spend very little time explaining these languages in class; at this point in your career you've been exposed to several programming language and are expected to be able to readily acquire new programming language skills.

If you are not sure whether this class is for you, please talk to me.

Required Text: Artificial Intelligence: A Modern Approach Second Edition (the green book) by Russell and Norvig, 2002.

Optional Text: Lisp Third Edition by Winston and Horn, 1989.

Readings. The textbook for the course is the second edition of Artificial Intelligence: A Modern Approach by Russell and Norvig. There are significiant differences between it and the first edition, so be sure to have the right edition. We will follow the textbook quite closely (although time will not permit us to cover all of the chapters), so it is imperative that you have a copy of the book. We may occasionally use supplemental readings as well, but those will be provided for you. The Lisp book is for your own reference in learning lisp, the language that most of your projects will be implemented in. There will be in class lessons on lisp as well as numerous web resources to assist you here, but the book can be invaluable.

Computing. You will have access to CoC clusters for your programming assignments. You can use whatever machines you want to do the work; however, the final result will have to run on the standard CoC boxes. Exactly what this means will be spelled out on each assignment. This shouldn't be much of a restriction for you.

• Allegro Common Lisp installed in the following labs:
  • Thin-Client Lab (CCB 130)
  • Baird Lab (CCB 107A)
  • Remote access unix machines (run alisp)
• Download your own:
  • Unix: Clisp (http://www.clisp.org/)
  • All platforms (limited heap size): Allegro Common Lisp (http://www.franz.com/downloads/)
  • Others

Web. We will use the class web page for most information, and T-Square for project submission. Aside from that, if you want to learn more about intelligent systems or artificial intelligence, you can find an enormous amount just by typing in keywords in google (or whatever your favorite search engine is). One good place to start is with AI on the Web. It's also worth pointing out that Georgia Tech enjoys one of the largest IS groups around and our interests are quite broad, so surfing faculty web pages can also be enlightening.

I reserve the right to modify any of these plans as need be during the course of the class; however, I won't do anything too drastic, and you'll be informed as far in advance as possible.

I expect you to understand and follow the honor code.