Computer Science, Prifysgol Cymru Aberystwyth University of Wales

C362(h)* - Artificial Intelligence Concepts

Brief Description

Artificial Intelligence (AI) has made many important contributions to computer science in general, and most experts believe AI techniques will become increasingly important. This module introduces students to the fundamental concepts of AI. Key issues including knowledge representation and reasoning, expert systems and learning are addressed both theoretically and practically.

Aims, Objectives, Syllabus, Booklist

Further Details

Number of lectures: 22
Number of seminars/tutorials: 0
Number of practicals: 6 x 1-hour
Coordinator: Dr. Fred Long
Other staff involved: Mr. Patrick Olivier
Pre-requisites: C210, C220
Co-requisites: None
Incompatibilities: None
Assessment: Assessed coursework - 20%
Written exam - 80%
Timing: This half module is offered only in Term 1

Aims

Artificial Intelligence is the study of computer systems which can perform the sort of tasks that are usually associated with human intelligence. Examples are: chess playing, pattern recognition, speech understanding and problem solving. The aim of this module is to introduce the main ideas and current problems in Artificial Intelligence including the key concepts of knowledge representation and reasoning, expert systems and learning. Students will be required to implement these concepts using an Artificial Intelligence programming language.

Objectives

On successful completion of this module students should:

understand the importance of knowledge representation and their role in Artificial Intelligence systems;
understand the use search in the solution of problems such as path planning and game playing;
be familiar with the principle techniques used in implementing machine learning;
understand the fundamentals of first generation expert system technology, and the conceptual basis of current attempts to overcome its limitations;
recognise the distinction between symbolic approaches to Artificial Intelligence and sub-symbolic approaches;
be capable of implementing fundamental Artificial Intelligence algorithms in the high level Artificial Intelligence language POP-11.

Syllabus

Introduction - 1 Lecture, 6 Practicals: The origins of Artificial Intelligence; 4 definitions of Artificial Intelligence. Characterising good and bad problem domains.
Introduction to POP-11 - 3 Lectures: Poplog environment, TEACH, HELP, basic POP-11, list processing, recursion, advanced features.
Knowledge Representation - 3 Lectures: Issues in knowledge representation: representation adequacy; inferential adequacy; inferential efficiency; acquisitional efficiency. Knowledge Representation formalisms: propositional and predicate logic, procedural representations; semantic nets; frame-based representations.
Search - 2 Lectures: The role of search in Artificial Intelligence. Search strategies: basic search; heuristic search; and game playing. Two workshops implementing depth- and breadth-first search, heuristic search and search space pruning.
Expert Systems - 2 Lectures: Characterising first generation expert systems. Expert system structure: knowledge representation; control strategies. Limitations of current expert system technology.
Learning - 3 Lectures: The role of learning in Artificial Intelligence. Symbollic approaches to learning: rote learning; Winston's arch learning program; version spaces and CYC.
Qualitative and Model-Based Reasoning - 4 Lectures: Techniques and methods used to provide second generation expert systems. Qualitative vs. quantitative representation and reasoning. Modelling physical domains and naive physics. Structure, behaviour, function and purpose. Models for simulation and diagnosis.
The Physical Symbol Hypothesis - 1 Lecture: Strong and weak Artificial Intelligence - philosophical interlude.
Neural Nets - 2 Lectures: Simulated neural networks. Hill climbing and back propagation. Characteristics of back propagation.
Genetic Algorithms - 1 Lecture: Chromosomes and hereditary traits; survival of the fittest; crossover; the rank method.

Booklist

It is considered essential to purchase the following

: E. Rich and K. Knight. Artificial Intelligence. McGraw Hill, 1991.

Students are likely to need ready access to the following

: Ramsay, Barrett, and Sloman. POP-11: A Practical Language for Artificial Intelligence. Ellis Horwood, 1985.
: Charniak and McDermott. Introduction to Artificial Intelligence. Addison Wesley, 1985.
: P. H. Winston. Artificial Intelligence. Addison Wesley, 3rd. edition, 1992.

The following should be consulted for different approaches or for further information

: S. C. Shapiro. Encyclopedia of Artificial Intelligence. Addison-Wesley, 1992.

Version 2.1

Syllabus

Nigel Hardy Departmental Advisor

nwh@aber.ac.uk

Dept of Computer Science, UW Aberystwyth (disclaimer)