CIS 587 Artificial Intelligence: Syllabus

SPRING 1996

Instructor: Dr. Giorgio P. Ingargiola
Office: Computer Activity Building, Room 1038
Phone: (215)204-6825
e-mail: ingargiola@cis.temple.edu

PREREQUISITES

CIS 203 or permission of Instructor. Some knowledge of first-order logic and of Lisp or of Prolog is desirable.

TEXT

Russell,S.,Norvig,P.:
Artificial Intelligence: A Modern Approach
Prentice-Hall, 1995(Recommended)

GRADING

Programs: 40%
Final: 40%
Midterm: 20%

You may choose, after agreement with instructor, to do a project with a report and presentation in place of the midterm and of selected homeworks.

DESCRIPTION

The purpose of this course is to give students an understanding of Artificial Intelligence methodologies, techniques, tools and results. Students will use at least one AI-language [Lisp, Prolog]. Students will learn the theoretical and conceptual components of this discipline and firm up their understanding by using AI and Expert System tools in home assignments. Interactions between Artificial Intelligence and other disciplines will be explored.

HOMEWORKS

You will be assigned a number of homeworks. Most homeworks will involve programming in Lisp. Unless otherwise specified, you are expected to work on your own.

EXAMS

The midterm is on the third tuesday of March. The final exam is on the last day of classes. The exams are written, closed book, with small but numerous questions on specific topics and techniques covered in the course.

OUTLINE

Overview of history and goals of AI: 1 week

Tentative definitions. Turing's test. Knowledge vs. Symbolic Level. Relations with other disciplines, from Philosophy, to Linguistics, to Engineering. Review of AI successes and failures.

State Spaces, Production Systems, and Search: 2 weeks

State Space representation of problems. Problem solving as search. Constraints. Definition and examples of Production Systems. Heuristic search techniques. Two person games.

Knowledge Representation Issues: 1 week

Procedural Knowledge Representation vs. Declarative Knowledge + Reasoning. Facts, General Assertions, Metaknowledge. The Frame Problem.

Using First-Order Logic for Knowledge Representation: 2 weeks

Propositional Logic: Semantics and Deduction. First Order Logic: Semantics and Deduction. Unification. Resolution-based theorem proving. Using theorem proving to answer questions about the truth of sentences or to identify individuals that satisfy complex constraints. Logic Programming.

Common Sense Reasoning : 0.5 week

Nonmonotonic reasoning and modal logics for nonmonotonic reasoning. How to deal with Agents and their Beliefs.

Weak Slot-and-Filler Structures: 0.5 week

Semantic Nets and Frames. Scripts for representing prototypical combinations of events and actions.

Rule-Based Systems: 1 week

Pattern-matching algorithms. The problem of Control in Rule Based Systems. The Rete Algorithm.

Planning: 1 week

Representing plans. Partial order planning. Planning applications.

Statistical Reasoning: 2 weeks

Use of Certainty Factors in Rule-Based Systems. Associating probabilities to assertions in first-order logic. Bayesian Networks. Fuzzy Logic.

Learning: 3 weeks

Learning to classify concepts using features of their instances. Learning a concept [Induction] from examples. Explanation-Based Learning. Version Spaces. Neural Nets with back propagation.

ingargiola@cis.temple.edu