Creative Commons License

Fall 2013
Knowledge-Based AI (CS 4635A/B– UG Section, CS7637– Grad Section)
Course Description

Ashok'Goel

CS 4635A/7637 Mondays' Wednesdays'and'Fridays,'9:05'– 9:55
Klaus'2456

Joshua'Jones

CS 4635B Mondays' Wednesdays'and'Fridays,'13:05'– 13:55
Klaus'1456

3 Credits

Course description for Knowledge Based AI (from the course catalog):

Basic Course Description:

Structured knowledge representations. Knowledge-based methods of
problem solving, planning, decision making, and learning.

CS 4635 and CS 7637 are the undergraduate and graduate sections,
respectively, of the same class. In either case, this is a "core"
course. It is also a challenging course, involving significant amount
of independent work including both readings and projects.

---------------------------------------------------------------------

Outline of the Course:

Unit 1: Structured Knowledge Representations - ~5-6 weeks
Semantic Networks
Production Rules
Frames
Scripts
Constraints
Logic

Unit 2: Knowledge-Based Reasoning and Learning - ~4-5 weeks
Planning
Learning
Classification
Diagnosis
Configuration

Unit 3: Advanced Topics - ~3-4 weeks
Case-Based Reasoning
Analogical Reasoning
Visual Reasoning
Meta-Reasoning
Semantic Web

Instructor will post day-by-day class schedule on the class site.

---------------------------------------------------------------------

Readings:

There is no textbook; instructor will provide handouts from:

Artificial Intelligence, Patrick Winston, 3rd edition.
Knowledge Systems, Mark Stefik.
Artificial Intelligence, Stuart Russell & Peter Norvig, 3rd edition
Recent review and research papers on selected topics.

---------------------------------------------------------------------

Class Format (please read this carefully):

We will use the format of a flipped class in which students read the
readings in advance of each class and the classes are used for quizzes,
exercises, discussions and short lectures
(see http://en.wikipedia.org/wiki/Flip_teaching).
We will assign readings for each class well in advance of the class, and
we will expect each student to have read the assigned readings before the
class.

Most classes will begin with a short video related to knowledge-based AI.
The video will be followed by a short quiz on the assigned readings for the
class. The quiz will be followed by a short lecture or a group exercise or
a class discussion (or some combination of these).

The quizzes will not be graded. However, they will count towards class attendance
and participation. We expect about 45 classes and about 35 quizzes during
the term. We will expect all students to take at least 30 quizzes in this
class.

---------------------------------------------------------------------

Class Notes:

Graduate students will take turns taking notes in the class and posting
class notes to the class site(s) on T-square. Each graduate student
may need to take notes for up to three classes.

---------------------------------------------------------------------

Projects:

There will be a series of five (design/programming) projects, with
later projects building on earlier ones. Each project will be about
two to three weeks long. Students may program in any "standard" programming
language (such as C++ or Java).

For each project, we will expect each student to turn in a design report
in addition to the program and the output for the project. The design
report will describe the software architecture, the knowledge representations,
the reasoning methods, and experiments with the programs. For each project,
we will post the best few programs and reports on the class site on T- square.

For each student, we will count the best four projects towards the grade.
Please note that the projects become progressively harder; indeed, the
fifth project is quite challenging. Thus, the best course for most students
would be to do the first four projects well.

---------------------------------------------------------------------

Examinations:

There will be a mid-term examination in early October and a final examination
in mid December.

---------------------------------------------------------------------

Grading:

All grades will be normalized (i.e., "curved"); the undergraduates and
graduate students will be graded on different scales.

For undergraduate students:
Mid-Term Examination: 15% of grade
Final Examinations: 35% of grade
Each project: 12.5% of grade
Class attendance/participation: 10%

For graduate students:
Mid-Term Examination: 12.5% of grade
Final Examination: 30% of grade
Each project: 12.5% of grade
Class attendance/participation: 10%
Class notes: 7.5% of grade

(I know the totals exceed 100. Good for you!)

---------------------------------------------------------------------

Instructor: Ashok K. Goel
Office: TSRB 219
Email: goel@cc.gatech.edu (best way to contact me)
Office Hours: MWF, 10:10-10:55 am

---------------------------------------------------------------------

Instructor: Joshua Jones
Office: CCB 260
Office Hours: MWF 12-12:55pm
Email: jkj@cc.gatech.edu

---------------------------------------------------------------------

GTA: Rochelle Lobo
Email: rlobo3@gatech.edu
Office Hours: Tuesdays 11:45 am - 2:45 pm

---------------------------------------------------------------------

GTA: Varun Thakkar
Email: vthakkar7@gatech.edu
Office Hours: Thursdays 11 am - 1:30 pm

---------------------------------------------------------------------

GTA: Gongbo Zhang
Office Hours: Fridays 2:15-5:15 pm
Email: gzhang64@gatech.edu

---------------------------------------------------------------------

Ashok Goel

August 16, 2013