Introduction to Genetic Programming2012HTFull
|
|
Course plan
Lectures
2 days intensive course
Recommended for
PhD students
The course was last given
New course
Goals
Genetic Programming (GP) is an important machine learning technique. It is a powerful, domain independent paradigm that can be used not only for optimization problems, but for invention and discovery. In this course, we will go from the motivations behind GP, to the inner workings of the algorithms to actual software that you can use out-of-the-box.
Prerequisites
MSc in mathematics or computer science/engineering.
Contents
Genetic Programming (GP)
Evolutionary Algorithms
Problem solving with GP
Software for GP
Organization
This is an intensive course, consisting of two days of lectures and practical
exercises. The participants will have the opportunity to do the exercises
afterwards and apply GP to their own research area.
Estimated own work is 8 to 16 hrs to complete. A selection of datasets is
provided, and the participants can chose one to apply GP techniques to. They
can either use off-the-shelf software, their own software or modifying
existing software. The task will involve finding suitable parameters for the
algorithm, performing the training and then evaluating the quality of the found
solutions. Assessment will be based on the quality of the scientific report,
which should be written in the style of a conference paper.
Literature
A Field Guide to Genetic Programming, by W. B. Langdon.
The e-book is available free of charge.
Lecturers
See the course plan.
Examiner
Dr Simon Harding (local examiner: Nahid Shahmehri)
Examination
Assessment will be by practical work. Students will be asked to implement software to solve a typical GP problem (e.g. time series prediction) and write a short, scientific report on their findings.
Credit
3 hp
Organized by
CUGS/ADIT
Comments
Registrations are binding.
Page responsible: Director of Graduate Studies