SubjectsSubjects(version: 902)
Course, academic year 2021/2022
  
Statistical Pattern Recognition - M445021
Title: Statistické rozpoznávání
Guaranteed by: Department of Computing and Control Engineering (445)
Actual: from 2020
Semester: winter
Points: winter s.:5
E-Credits: winter s.:5
Examination process: winter s.:
Hours per week, examination: winter s.:2/2 [hours/week]
Capacity: unlimited / unlimited (unknown)
Min. number of students: unlimited
Language: Czech
Teaching methods: full-time
Level:  
For type: Master's (post-Bachelor)
Note: course can be enrolled in outside the study plan
enabled for web enrollment
Guarantor: Kukal Jaromír doc. Ing. Ph.D.
Interchangeability : N445088
Annotation -
Last update: Pátková Vlasta (20.04.2018)
The lectures are oriented to application of mathematical statistics in pattern recognition. Basic terms, their relationships to statistics, programming techniques, and descriptor design are poined out together with research, real data processing, and applications.
Aim of the course -
Last update: Pátková Vlasta (20.04.2018)

Students will be able to:

Design descriptors for statistical pattern recognition

Apply data transforms for improvement of pattern recognition quality

Apply standard pattern recognition methods

Evaluate quality of pattern recognition systém

Literature -
Last update: Pátková Vlasta (20.04.2018)

R:Fukunaga K., Introduction to Statistical Pattern Recognition, Academic Press, London, 1990

R:Shawe-Taylor J., Cristianini N., Kernel Methods for Pattern Analysis, Cambridge University Press, Cambridge, 2009

A:Scholkopf B., Smola A.J., Learning with Kernels, MIT Press, Cambridge, 2002

Learning resources -
Last update: Pátková Vlasta (20.04.2018)

interal training materials

Syllabus -
Last update: Pátková Vlasta (20.04.2018)

1. Class, pattern, descriptors, pattern set for statistical pattern recognition

2. Repetitorium of basic statistical terms and principles

3. Quality of desroptor set: the best descriptor, acceptable subset of descriptors

4. Linear discrimination analysis as a tool for pattern recognition

5. Evaluation of pattern recognition quality: p-value, sensitivity, specificity, error, AIC, BIC

6. Cross-validation methodology in evaluation of recognition quality

7. Linear data transforms: normaliztion, standardization, PCA, spherization

8. Robust and regularized methods and their advantages in pattern recognition

9. Application of metrics in pattern recognition: Euclidean, Minkowski, Mahalanobis, k-NN, c-mean

10. Application of PDF in pattern recognition: GMM, Parzen and LQ estimates

11. Linear, non-linear, and logistic regression as pattern recognition tools

12. Feature space reduction as application of binary optimization

13. Kernel functions in design of non-linear classifiers

14. Fuzzy sets in pattern recognition: fuzzification, FCM

Entry requirements -
Last update: Pátková Vlasta (20.04.2018)

Basic knowledge of mathematical statistics and programming in MATLAB.

Registration requirements -
Last update: Pátková Vlasta (20.04.2018)

none

Course completion requirements - Czech
Last update: Mareš Jan doc. Ing. Ph.D. (26.04.2018)

Vypracování a obhajoba tří samostatných programů: 0 - 25 bodů

Ústní zkouška: 0-75 bodů

Celkové bodové hodnocení: 100-90 A, 89-80 B, 79-70 C, 69-60 D, 59-50 E, méně než 50 F.

Teaching methods
Activity Credits Hours
Konzultace s vyučujícími 0,5 14
Účast na přednáškách 1 28
Práce na individuálním projektu 2 56
Příprava na zkoušku a její absolvování 0,5 14
Účast na seminářích 1 28
5 / 5 140 / 140
 
VŠCHT Praha