8002303 Pattern Recognition, Spring 2006, 5 credits



NEWS


6.04. There will be no lecture at that day at 12-14. This lecture is shifted to 20.04 at 14-16. Respectively, the group participating the exercises on thursday at 14-16 will have no exercise class on 20.04. This exercise is shifted to 6.04: at 12-14 in the room TC407 (instead of the lecture). The exercise on 6.04. at 14-16 will be given according to the schedule.
11.04. The lecture at that day is shifted from 12-14 to 14-16. The exercise group participating the exercises on tuesdays at 14-16 will have their exercise at 12-14 at lecture time in the room TC 415.
12.04.-18.04. Easter holidays. No lectures and exercises at that days.
 
NOTICE: The exam will be graded with numeric assesment (1-5).
 
  Lecture notes available for copying in the pigeon hole in Tietotalo, 4th floor.



Lectures (24 h):

Lectures are given by Prof. Ulla Ruotsalainen and Prof. Ari Visa.
First lecture: 28.03.2006
Time and Place: Tuesdays 12-14, TB 219
Thursdays 12-14, TB223


Exercises (12h):

Time and place:
1 group Tuesdays 14-16, room TC415
2 group Tuesdays 16-18, room TC415
3 group Thursdays 14-16, room TC415
You have to choose one exercise group at time most suitable to you and participate it.
First exercise: 28.03 and 30.03.


Prerequisites:

SGN-2506 Introduction to Pattern Recognition (Lectures)
SGN-2500 Johdatus hahmontunnistukseen (Luennot)


Requirements:

Written exam, set of exercises and two small homeworks. In order to pass the course the exercises and the homeworks should be completed. After each session you have to write a short report (2-3 pages long) according to the following template .
The assistants for the exercises are: Evgeny Krestyannikov and Carol Rus. Exercises 1-3 will be given by Evgeny Krestyannikov, exercises 4-6 by Carol Rus. Please, return the reports by email to the respective assistants within one week after each exercise session.

Tasks:

Exercise 1: Bayes classifier. Matlab files
Exercise 2: Linear classifiers (perceptron,LMS,MSE). Matlab files
Homework 1: Multicategory classifiers. Matlab files
Exercise 3: Stochastic classifiers (Boltzmann machine). Matlab files
Exercise 4: Tree classifiers. Matlab files (updated)
Homework 2 (Due to 12.05.2006): Algorithm-independent machine learning. Matlab files
Exercise 5: Projections and feature extraction. Component and discriminant analysis. Matlab files
Exercise 6: Unsupervised learning and clustering (k-means,hierarchical, probabilistic). Independent component analysis. Matlab files

Accepted reports:

List of accepted reports can be found here

Literature:

R.Duda, P.Hart and D.Stork: Pattern Classification, 2nd edition, Wiley-Interscience, New-York,2001(Powerpoint lecture slides)


The following chapters will be be covered during the course:
Chapter 2: Bayesian decision theory and belief networks
Chapter 3: Bayesian parameter estimation, sufficient statistic, Expectation-Maximization method, Hidden Markov Models
Chapter 5: The whole chapter
Chapter 7: The whole chapter
Chapter 8: The whole chapter
Chapter 9: The whole chapter
Chapter 10: Repetition of mixture densities, clustering, component analyses(PCA,ICA,NLCA,Factor analysis)
In addition: Fuzzy methods from all chapters.