Swedish web site
Introduction to machine learning and its applicationsFDA178, 2005VTFull
|
|
Course plan
No of lectures
Probably as follows:
Part 1: 6-7 3-hour sessions (18-21 hours in total).
Part 2: ~3 3-hour sessions (9 hours in total, but will depend on the number of
participants).
Recommended for
PhD students in computer science.
The course was last given
The course is loosely based on a study circle at ADIT given in 1998. Other related graduate courses previously taught at IDA include FDA140, FDA133, and FDA093.
Goals
There are two main goals:
1. To provide an overview of fundamental machine learning techniques, and
2. To provide opportunities for in-depth study of how these techniques can be
(or have been) applied to practical learning problems.
Prerequisites
Organization
The course is organised as a combination of lectures and study circle,
and it consists of two parts.
Part one will start with a brief overview lecture introducing the topic of
machine learning. Afterwards, participants will in turn study and present
fundamental machine learning techniques (see content description below).
In part two, each participant will choose (or suggest) a topic for in-depth
study. This in-depth study will investigate how machine learning techniques can
be applied to a certain type of learning problem (e.g. for anomaly detection,
user profiling, etc). This study could be organised as a literature survey or
as an implementation project (or a combination). These studies will end with a
written report and presentation.
Contents
Fundamental machine learning topics include concept learning, decision
tree learning, artificial neural networks, Bayesian learning, instance-based
learning, genetic algorithms, and reinforcement learning.
Based on the interest of the participants, there may also be possibilities to
include more recent developments (e.g. ensemble learning, support vector
machines) to this list of fundamental topics.
Literature
The following book by Tom Mitchell will be used to cover the fundamental topics
studied in part one.
Tom Mitchell, 'Machine Learning', McGraw-Hill, 1997.
In addition, other material may be used to cover more recent advances
in machine learning.
Part two will be based mainly on research papers, and will depend on the type
of project suggested by the participant.
Lecturers
Johan Åberg, and possibly guest lecturers.
Examiner
Nahid Shahmehri and Johan Åberg
Examination
Part one: active participation and presentations.
Part two: project report and presentation.
Credit
3+3 credits.
Comments
Will likely be given in the later part of VT2005.
Page responsible: Director of Graduate Studies
