Machine Learning represents the study of computer algorithms that improve automatically through
'experience'. Genetic algorithms/ programming is a biologically motivated method for manipulating a
?population? of algorithms/ programs such that they improve as they experience an environment. The
Genetic Programming approach, however, is but one of a family of search methods, referred to as
Evolutionary Computation. This course focuses on Genetic Algorithms and Genetic Programming alone
with the basic objective of providing a thorough grounding in this family.
Course Format
The course consists of a combination of lectures, student presentations, coursework and
project.
Evaluation
Course evaluation is conducted on the basis of presentations (20%), coursework assignments
(15%),
and project (65%).
Presentations are considered a group activity. Coursework and project are performed individually.
All three activities are subject to the university plagiarism policy, where it is the responsibility of
the student to be familiar with this!
Coursework is typically employed to establish uniform knowledge of the background material.
Project work is typically - but not necessarily - application orientated and provides you with the opportunity
to experiment with the
concepts visited during the course. Periodic presentations to class are expected at 4 points in the term in
order to minimize the expectation of a miricle at the end of term. Marks for your project however are all derived
from your final project report. This takes the form of a 'conference paper' presented to your peers at the end of
term.
Marking scheme for your project might take the following form, however, depending on your chosen topic may
be modified as appropriate:
Generic Marking Scheme for a GP application type project,
Representation - 10%
Selection Operators - 10%
Cost function - 10%
Experimental framework - 20%
Code bloat/ Coevolution - 10%
Relation to previous works - 5%
All decisions regarding coding are the responsibility of the student. You are free to
make use of open source code, however, all modifications to support your
project are naturally your responsibility.
Readings
In addition to the Course Text, the course will be supplemented with material from conferences and
journals as required. Moreover, additional reading may be conducted/
necessary on a case-by-case basis.
Melanie Mitchell, An Introduction to Genetic Algorithms. MIT Press, 1996.
Course Outline
A general guide and will vary on a year-to-year basis, see course homepage
for a complete set of potential lecture topics.