Laboratory of Computer and Information Science

Tik-61.184 Informaatiotekniikan erikoiskurssi IV (4 ov) (L)

What's New
24.9.	Instructions for Project work.
21.9.	Timetable: topics assigned to people.
15.9.	Book order (DL 21.9.)
	Timetable:indicate topics that you prefer (DL 17.9.)

NOTICE: The time has changed: Tuesdays 14-16 at B333

Lecturer: Docent Samuel Kaski

Assistant: M.Sc.(Tech) Krista Lagus, krista.lagus@hut.fi, room C311
Semester: Autumn 1999
Credit points: 4 cr
Place: Seminar room B333 in the computer science building
Time: Tuesdays 14-16, starting from September 14
Language: English
Homepage: http://www.cis.hut.fi/cis/edu/Tik-61.184

Self-Organizing Maps

The Self-Organizing Map (SOM) is a widely used neural network algorithm that is especially suitable for visualizing and interpreting large high-dimensional data sets. Already over 3700 scientific publications have been written on the SOM.

The SOMs are also known as "Kohonen Maps" after their inventor, Academy Professor Teuvo Kohonen, who will give an opening lecture in the beginning of the course (Sept. 14, 14:15). The SOM is one of the main research topics of the Neural Networks Research Centre and the Laboratory of Computer and Information Science; for more information see http://www.cis.hut.fi/.

The seminar covers the state of the art of SOM research and applications. It is based on articles from the top researchers on SOM, collected in the new book "Kohonen Maps" (Editors: E. Oja & S. Kaski, Elsevier, Amsterdam, 1999). More details of the book can be found on the Web page of the course. There will be some books available in the laboratory. Additionally, the book "Self-Organizing Maps" (T. Kohonen, Springer, Berlin, 2nd Ed. 1997) is recommended although not necessary for the seminar. It is by far the most comprehensive treatment of SOMs.

Purchasing the books:

Kohonen Maps: Ask secretary Tarja Pihamaa (tel. 451 3272, tarja.pihamaa@hut.fi, room B326)
Self-Organizing Maps: A common order will be arranged. You can still sign for the book order in next Tuesday's session (21.9.).

Requirements for passing the course:

Give a presentation of some chapter of the book
Solve a set of exercises (at least 60% of given ones)
Make a project work
Participate actively in the seminar

To pass "with distinction", at least 95% of the exercises should be solved, and the presentation as well as the project work should be very good.

Presentations

There will be approximately two presentations each week. Each presentation should last about 35 minutes, so that 10 minutes is left for questions and discussion.

Your task is to present the paper in such a way that the audience can understand the topic based on your presentation (at least in principle...) Some words of advice that may be useful:

The topics are not easy; some background material may really be needed in undestanding the papers. Therefore it may be a good idea to start preparing early enough.
Prepare clear overheads and try to be clearer than the original paper!

Timetable

Date	Speaker	Topic	Start page
14.9.		Introduction to SOM by prof. Teuvo Kohonen; practical arrangements about the course.
21.9.	Simona Malaroiu	Value maps: Finding value in markets that are expensive	15
21.9.	Atte Saarela	A SOM-based sensing approach to robotic manipulation tasks	207
28.9.	Markus Varsta	Analyzing and representing multidimensional quantitative and qualitative data:Demographic study of the Rhone valley. The domestic consumption of the Canadian families.	1
28.9.	Razvan Cristescu	Active learning in Self-Organizing Maps	57
5.10.	Johan Himberg	Self-Organizing Maps on non-Euclidean spaces	97
5.10.	Markus Koskela	Tree structured Self-Organizing Maps	121
12.10.	Ville Viitaniemi	Growing self-organizing networks --- history, status quo, and perspectives	131
12.10.	Olli-Pekka Rinta-Koski	On the optimization of Self-Organizing Maps by genetic algorithms	157
19.10.	Henri Hakonen	Self organization of a massive text document collection	171
19.10.	Mikko Syrjälahti	Document classification with Self-Organizing Maps	183
26.10.	Jarmo Ritola	Self-Organising Maps in computer aided design of electronic circuits	231
26.10.	Timo Mäkelä	Modeling self-organization in the visual cortex	243
2.11.	Juha Raitio	Topology preservation in Self-Organizing Maps	279
2.11.	Jussi Ahola	LVQ and single trial EEG classification	317
9.11.	Vuokko Vuori	Chemometric analyses with Self Organising Feature Maps: A worked example of the analysis of cosmetics using Raman spectroscopy	335
9.11.	Tuomas Pantsar	Self-Organizing Maps for content-based image database retrieval	349
16.11.	Klaus Riederer	Indexing audio documents by using latent semantic analysis and SOM	363
16.11.	Antti Saarela	Self-Organizing Map in analysis of large-scale industrial systems	375
23.11.	Tommi Noponen	A spatio-temporal memory based on SOMs with activity diffusion	253
30.11.		Feedback seminar for project work; enroll by 16.11. by sending email to course assistant.

Project work

The purpose of the project work is to get hands-on experience with applying the SOM to data analysis, as well as on reporting your work to peers. The deadline for the project work is 31.1.2000.

See instructions for the project work.

Exercises

Paper copies of exercises will be available in the seminar sessions. Solved exercises may be returned either during seminar sessions or to the course assistant (in room C311). The deadline for solving the exercises is 31.1.2000.

Other CIS courses
September 21th, 1999
Krista Lagus
Samuel Kaski