Prerequisite Modules
Description
Artificial intelligence is the area
of computer science that treats the advanced algorithms required to
make a computer behave intelligently. This intelligence covers a wide
spectrum from models of mind in artificial agents to industrial
applications that record preferences and make recommendations. The
philosophical and psychological foundations of the area are necessary
for a deep understanding of the potential and implications of the
study of artificial intelligence (AI). Traditionally, AI research has
been divided into two competing fields - the symbolic domain where
symbol processing, manipulation and inference are treated, and the
subsymbolic domain where statistical and connectionist methods are
treated. This module provides the student with an overview of the
entire field, coupled with an in-depth treatment of some of the
advanced algorithms and aspects of the two sub-domains into which AI
is divided.
Aims
-
To provide a foundation in the major
elements of Artificial Intelligence.
-
To introduce different knowledge
representation techniques and discuss the issues involved.
-
To establish the basis for
mechanical inference.
-
To introduce connectionist and
neural architectures.
Learning Outcomes
On successful completion of this
module, the student will be able to:
-
Identify the major issues of the
wide spectrum of AI.
-
Offer comparative appraisal of
accepted definitions of AI.
-
Discuss potential applications of
intelligent techniques.
-
Demonstrate an understanding of
the major knowledge representation schemes and inference mechanisms
based on these schemes.
-
Represent knowledge in logic
formalisms and to apply standard proof techniques.
-
Demonstrate an appreciation of a
number of classical optimisation problems and the difficulties of
solving these.
-
Describe and use genetic
algorithmic techniques for problem solving.
-
Compare and contrast the
capabilities of the two main areas of AI.
-
Describe and implement a number of
connectionist architectures.
Learning and Teaching Methods
Lectures, self-study, labs,
tutorials, and any combination of discussion, case study,
problem-solving exercises, readings, seminars, and computer-based
learning.
Content
Foundations of Artificial
Intelligence: Philosophical foundations.
Psychological foundations. Adaptive behaviour. Learning. Agency.
Theorem Proving:
The Role of Logic, Inference. Unification, Role in Theorem Proving,
Legal Substitutions, Algorithms. Deduction and Term Rewriting, Modus
Ponens, Modus Tolens, Reduction, Applications. Resolution, Clause
Form, Proof By Contradiction, Techniques, Applications. Query
optimisation.
Strong method problem solving:
Production rules Forward and Backward chaining. Reasoning with
Uncertainty, Probalistic reasoning, Certainty Theory, Fuzzy Logic.
Optimization Problems and
Decision Problems: Approximation Algorithms.
Bin Packing. The First Fit Decreasing Strategy. Other Heuristics. The
Knapsack and Subset-Sum Problems Graph Coloring. Some Basic
Techniques. Approximate Graph Coloring Is Hard.Wigderson's Graph
Coloring Algorithm. The Traveling Salesperson Problem Greedy
Strategies. Simulated Annealing The Nearest Neighbor Strategy. The
Shortest Link Strategy. TSP Approximation Algorithms?
Genetic Algorithms:
Computing with DNA DNA Background. Adleman's Directed Graph and the
DNA Algorithm. Analysis and Evaluation.
Neural Architectures:
Backpropogation, Associative Networks and Pattern Recognition,
Bidirectional Associative Memories(BAM)s, Hopfield Model Learning in
Hopfield Networks, A Thermodynamic analogy, The Boltzmann Machine, The
Kohonen Model,. Self Adaptive Topological maps, Specialised neurons,
Lateral Interaction Between Neurons.
Assessment
The methods of assessment to be
used to measure the learning objectives stated above are written
examination and continuous assessment including one or more of
assignment, essay, problem-solving exercise, oral presentation, and
class or lab tests.
-
Continuous Assessment: 30%
-
Examination: 70%
Recommended
Reading
-
George F Luger (2002), Artificial
Intelligence: Structures And Strategies for Complex Problem Solving
(4th Edition), Addison Wesley
-
Stuart Russell and Peter Norvig
(1995), Artificial Intelligence A Modern Approach, Prentice Hall
-
Kevin Gurney (1997), Introduction
to Neural Networks, Routledge
-
Adaptive Behavior, Journal of The
International Society for Adaptive Behavior
http://isab.org/journal/
-
Autonomous Agents and Multi-Agent
Systems, The official journal of the International Foundation for
Multi-Agent Systems (IFMAS),
http://springeronline.com/
|
|
For more information contact
Ciarán O'Leary
|
Home