Pattern Recognition

Course Description

Pattern recognition is the scientific discipline whose goal is the classification of objects into a number of categories or classes. These objects can be images (2D signals) or signal waveforms (1D signals) or any type of measurements that need to be classified. The objects are refered using the generic term patterns. Pattern recognition is an integral part of machine intelligence systems.

General Competencies

The course "Pattern Recognition” enables the students to understand basic, as well as advanced techniques of pattern classification and analysis that are used in machine interpretation of a world and environment in which machine works. Pattern recognition is basic building block of understanding human-machine interaction.

Learning Outcomes

  1. explain and define concepts of pattern recognition
  2. explain and distinguish porocedures, methods and algorithms related to pattern recognition
  3. apply methods from the pattern recognition for new complex applications
  4. analyze and breakdown problem related to the complex pattern recognition system
  5. design and develop a pattern recognition system for the specific application
  6. evaluate quality of solution of the pattern recognition system

Forms of Teaching

Lectures

Classes are held in two phases - each 7 weeks. Classes are conducted over 15 weeks with a weekly load of three hours. After each phase, ie, in the 8th week of lectures and 15th week of lectures exames are held.Week immediately prior to the exams is scheduled for problem solving and illustrations of procedures.

Exams

Knowledge checking is done by written examination twice in a semester.

Consultations

Consultations are planned for 2 hours per week.

Seminars

Groups of 4 to 7 students receive project tasks. The group solves problem, implements the pattern recognition system and evaluate it.

Grading Method

Continuous Assessment Exam
Type Threshold Percent of Grade Threshold Percent of Grade
Seminar/Project 50 % 20 % 0 % 20 %
Mid Term Exam: Written 50 % 40 % 0 %
Final Exam: Written 50 % 40 %
Exam: Written 50 % 80 %

Week by Week Schedule

  1. Task of pattern recognition. Features, Feature vectors, Classifier.
  2. Pattern Recognition System Model.
  3. Determining linear decision functions. Gradient learning procedures. Perceptron algorithm with fixed correction. Variants of perceptron algorithm.
  4. Ho-Kashyap method. Generalized Perceptron Algorithm.
  5. Illustration of procedures. Numerical problem solving.
  6. Fisher’s linear discriminant analysis (LDA). Multiple discriminant analysis. Learning and testing sample sets – methods of evaluation.
  7. Illustration of procedures. Numerical problem solving.
  8. Midterm exam.
  9. Support Vector Machines (SVM).
  10. Nonlinear Classifiers. Generalized Linear Decision Functions. Dichotomy. Kernel Based Methods.
  11. Classifier Based on Bayes Decision Theory.
  12. Karhunen – Loeve transform (principal component analysis).
  13. Artificial neural networks. Three- , Multi-layer Perceptrons. Error-backpropagation Learning Algorithm. Hidden Markov Models.
  14. Illustration of procedures. Numerical problem solving.
  15. Final exam.

Study Programmes

University graduate
Computer Science (profile)
Theoretical Course (1. semester)
Information Processing (profile)
Theoretical Course (1. semester)

Literature

S. Theodoridis, K. Koutroumbas (2009.), Pattern Recogniton, Elsevier
R.O. Duda, P. E. Hart, D.G. Stork (2001.), Pattern Classification, J. Wiley, New York
L. Gyrgyek, N. Pavešić, S. Ribarić (1988.), Uvod u raspoznavanje uzoraka, Tehnička knjiga Zagreb
L. I. Kuncheva (2004.), Combining Pattern Classifiers: Methods and Algorithms, Wiley-Blackwell

Lecturers

Grading System

ID 34404
  Winter semester
5 ECTS
L1 English Level
L1 e-Learning
45 Lecturers
0 Exercises
0 Laboratory exercises

General

89 Excellent
74 Very Good
61 Good
50 Acceptable