SGN-2506 Introduction to Pattern Recognition, 4 cr., fall 2011

Last update: Dec 17, 2011

News

Dec 12, The final record of exercise points is published (see links below). Note that now the record shows your percentage of the maximum instead of points.

Dec 8, The lecture slides are available for download as a zip file .

Nov 4, 2011 Collected exercise points after the first exercise can be found here . If you think there is an error, please contact Mikko Parviainen by email. Oct 26, 2011 The sign-up is now open. It will be closed on Wed 2, 2011 at 09:30 o'clock.

Oct 20, 2011 Sign-up for exercises will be opened via POP. Choose either Wednesday's or Thursday's exercise group.

Oct 20, 2011 The first exercise is on November 2 (Group 1) and November 3 (Group 2).

Oct 19, 2010 Page updated for 2010 edition. Check this page regularly for further updates. -->

[Lectures] [Exercises] [Contents] [Requirements] [Exams] [Literature] [Links]

Lectures (28h):

On Wednesdays 12 - 14 and on Thursdays 10 - 12, Room TB223
Lectures: Jari Niemi, jari.a.niemi(<-at->)tut.fi

Lectures are in English.

Material

Handout

Course schedule

• 26 Oct Introduction
• 27 Oct Basics on probability and statistics
• 2 Nov Basics on probability and statistics
• 3 Nov Bayesian decision theory
• 9 Nov Bayesian decision theory
• 10 Nov Maximum likelihood classifier (parameter estimator)
• 16 Nov Parzen classifier
• 17 Nov Parzen classifier, probabilistic neural network
• 23 Nov K-nearest neighbors classifier
• 24 Nov Linear classifier
• 30 Nov Linear classifier, classifier evaluation
• 1 Dec K-means classifier (clustering)
• 7 Dec Mixture model classifier (clustering)
• 8 Dec Recapitulation, examples of earlier exams

Exercises (12h):

Exercise arrangements: Mikko Parviainen (mikko.p.parviainen(<-at->)tut.fi).

Exercises start on week 44 (one week after the first lectures).

There six exercise sessions in total. Each session lasts approximately two hours.

Exercise Groups:

Wed 10 - 12, room TC131

Thu 12 - 14, room TC407

Requirements:

The exercise problems are individual work for each student. Sharing of answers is not allowed. At the beginning of each exercise session, a list will be given, on which you can mark the problems you have thought over and you are ready to write on the board.
At least 30 % of the exercises must be performed in order to be able to attend the exams. The 30 % threshold is not flexible. Bonus points are awarded as follows: at least 40%: 1 point, at least 60%: 2 points, at least 80%: 3 points, at least 90%: 4 points. The bonus points are taken into account in the three first exams of the course. Notice that the exercise points as well as the correct answers you can get only by attending the exercise sessions.

Exercise problems

Exercise 1

Exercise 2

Exercise 3

Exercise 4

Exercise 5

Exercise 6

Exercise points

Goal and Contents

The goal is to introduce basic methods and principles of pattern recognition. Basics of multivariate probability and statistics. Bayesian decision theory. Parameter estimation from training data. Non-parametric techniques for pattern classification. Algorithms for unsupervised classification.

Requirements:

Final examination and active participation in exercises (30 % minimum).

Exams:

Visit POP for dates and sign-up (required).

Literature:

Tohka: SGN-2506: Introduction to Pattern Recognition, 2006 - 2008 (revised on Oct 13, 2011).
Duda, Hart, Stork: Pattern Classification, 2nd edition, Wiley, 2001.

Links:

Images of the course book

Pattern Recognition (Wikipedia)

Feature selection and clustering by R. O. Duda

Bilmes: A Gentle Tutorial of the EM Algorithm and its Applications to Parameter Estimation for Gaussian Mixture and Hidden Markov Models

HY - Tilastollinen hahmontunnistus