Computer Science 687 / 001

Advanced Artificial Intelligence

Meets

Monday, 4:30–7:10 PM, in East Building 122.

Professor

Sean Luke.

Prerequisite

CS480, CS580, or an equivalent class taught at another university. This is a hard prerequisite: generally students who have not met this prerequisite will fail CS687 and become sad.

About the Class

This course will cover several advanced topics in Artificial Intelligence beyond those covered in CS580. These topics will extend existing knowledge about search, machine learning, reasoning, and situated action. Some topics are required; others may be negotiated with the class. Topics may include planning, probabilistic reasoning, reinforcement learning, evolutionary computation, advanced neural networks, natural language processing, constraint satisfaction, reactive systems, knowledge-based learning, robotics, vision, emergent behavior, and intelligent multiagent systems.

AI is a very broad field, and the goal of this course is to provide the student with sufficient breadth beyond CS580 to act as a well-versed AI researcher. Informally, CS580 + CS687 try to cover all the topics that might show up on the Ph.D. Qualifying Exam.

All programming assignments will be done in Common Lisp, the traditional exploratory programming language of AI. Assignments will also be done by teams of students. Often but not always Common Lisp is taught in CS580, but some students may not have learned it, so we'll have to teach it.

Course Web Page

http://cs.gmu.edu/~sean/cs687/      This class does not use Blackboard or Piazza.

Texts

Artificial Intelligence: A Modern Approach SECOND (green) or THIRD (blue) editions, by Russell and Norvig. Do not get the FIRST (red) edition.
ANSI Common Lisp by Paul Graham, ISBN: 0133708756. This book is not required but is recommended. Amazon has many cheap used copies.

These textbooks are readily bought on Amazon, Barnes & Noble, etc., and you do not need them in the first week or two.

Grading

This course will consist largely of several large projects and two exams. The breakdown will be approximately: 1. Homework and (several) Projects: 50% with higher weight given to harder projects. 2.Exams (2 of them): 25% Each

Course Outcomes

1. An ability to employ various basic machine learning, probabilistic reasoning, search, knowledge representation, and planning techniques. 2. An ability to use Lisp at a moderate to high level of proficiency. 3. An ability to develop nontrivial artificial intellience applications. 4. An ability to work in teams. 5. An ability to identify the right AI technique to use for a given problem, and to understand the issues and tradeoffs involved.