Pattern Recognition

The course is a substitute for:

Pattern Recognition (XE33RPZ)

Lecturer:

Tutor:

Supervisor:

Department of Cybernetics

Synopsis:

The course gives an introduction to statistical and syntactic pattern recognition. The classification (pattern recognition) problem as risk minimization. Bayes decision-making. Parametric classifiers. Learning. Parameter estimation. Non- parametric classifiers. The nearest neighbour method. Neural nets principles. and learning. Testing. Feature selection. Support Vector Machines. Cluster analysis. Structural risk minimization. Syntactic Pattern Recognition. Languages, grammars, automata. Parsing, syntactic classification. Applications of Pattern Recognition.

Requirements:

Syllabus of lectures:

1. Pattern recognition problem formulation. Basic terms, a map of the course. Application examples

2. Bayesian formulation of the decision problem as a minimalization of the expected loss

3. Non-Bayesian statistical pattern recognition problems

4. Parameter estimation. The maximum likelihood method. Bayesian estimation 5. EM algorithm (expectation maximization)

6. Non-parametric methods. Nearest neighbour method. Parzen windows

7. Classifiers assuming normal distribution of features. Linear discrimination function

8. Neural nets 1. Perceptron. The perceptron algorithm and its convergence

9. Neural nets 2. Multilayer networks MLP (multilayer perceptron) and RBF (radial basis functions). Learning by back-propagation

10.Vapnik´s theory of learning. Structural risk

11. Support vector machines

12. Feature selection

13. Cluster analysis

Syllabus of tutorials:

1. - 14. A group of two students designs a classifier for character recognition. The data used are from a large, publicly available database. In the first part a simple classifier is trained on a relatively small number of samples. In a second part of the experiment the students explain the differences between a predicted and actual test error on an independent test set.

Study Objective:

Study materials:

[1] Devijver, Kittler: Pattern Recognition. Prentice Hall, 1982

Note:

Further information:

No time-table has been prepared for this course

The course is a part of the following study plans: