Advanced Pattern Recognition (Spring 2007), 173353, 5 ects

Goal

To give understanding to the modern kernel methods in pattern recognition and to give an idea about different applications. Also the student will get practical skills to use some kernel methods in a pattern recognition task.

>>>>>> Exam results

Teaching arrangements and studies

Teachers: Jussi Parkkinen, Alexander Kolesnikov
Guest Lecturers: Prof. V.Shkodyrev (Saint-Petersburg), Prof. H.Kälviäinen (Lappeenranta).

In English, 32 h
Wednesday 14-16 and Thursday 12-14 beginning from 07.03.2007.
in Science Park, lecture room 2D106.

20.03--23.03.2007: Short intensive on Neurinformatics (8 h) Course by Prof. V.Shkodyrev, Saint-Petersburg State Polytechnical University, Russia.

1) 20.03, Tuesday, APR demo time, 12-14, B181
2) 20.03, Tuesday, 16-18, 2D106 (preliminary time)
3) 21.03, Wednesday, APR lecture time: 14-16, 2D106
4) 22.03, Thursday, APR lecture time: 12-14, 2D106

28.03, Wednesday, APR projects and Introduction into Matlab, Alexander Kolesnikov
29.03, Thursday, APR lecture, Jussi Parkkinen
18.04, Wednesday, short lecture (On-line classification algorithm) and projects discussion, Alexander Kolesnikov

26.04, 12:14.00, Thursday, Guest Lecture, Prof. H.Kälviäinen

Lecture Notes

  1. Lecture 1
  2. Lecture 2
  3. To Fisher Discriminant Analysis
  4. Principal Component Analysis
  5. Kernel PCA
  6. On-line classification algorithm

Neuroinformatics by Prof.Shkodyrev

Excercises

Tuesday 12-14
in Science Park, B181
Link to Demo

Deadline for projects: 03.05.2007.

Final exam on APR: Thursday, 10.05.2007 at 12.00-14.00 ?
We can discuss the date on Wednesday, 2.05.2007, at 14.15 (APR lecture time).
Questions for the exam: kernal methods for Pattern Recognition according to the list of projects.

Links

  1. Kernel Methods for Pattern Analysis by John Shawe-Taylor & Nello Cristianini, Cambridge University Press, 2004.

  2. Kernel Machines Website
    • Publications
    • Tutorials
    • etc.

  3. Introduction to Kernel Methods by Bernhard Scholkopf (slides).