Warning

This page is located in archive.

The subject introduces pattern recognition (called also machine learning). The subject is intended for doctoral students. I can take master students after an initial interview about student's motivation as well. The lectures are open to bachelor students but no credit for the subject will be awarded.

It is expected that student did not study this subject before, e.g. in the bachelor subject Rozpoznávání a strojové učení, B4M33RPZ, B4M33RPZ lectured by Prof. Jiří Matas. The lecturer will ask the student if she/he did not pass a similar subject to avoid getting credits easily.

The subject comprises of lectures and consultations with the lecturer (if the student asks for it).

It is required that the student passed an introductory course of the probability theory and statistics or studied this area individually.

- The lectures are scheduled regularly for Wednesdays from 9:15 to 10:45 in JP:B-633. We will discuss the date/time of the lecture with the students at the first lecture. We might find a better date/time that suits the majority of students and the lecturer.
- The second lecturer is Mgr. Radoslav Škoviera, Ph.D.
- 2 x 45 minutes of lecture a week, i.e. 90 minutes in one single lecture.
- If there is one non-Czech-speaking student in the audience then the lecture will be given in English. The lecture presentation will be in English typically even if the lecture is given in Czech.
- Irregularities in the lecture schedule might be needed due to the lecturers' business trips. The replacement date will be discussed with the students in advance.
- Lecturer's abbreviation VH - Václav Hlaváč, RŠ - Radoslav Škoviera

Week | Date | Time | Who | Content |
---|---|---|---|---|

01 | 16.02.2022 | 09:15 | VH | Outline of pattern recognition, Intro to pattern recognition |

02 | 23.02.2022 | 09:15 | VH | Rehearsal of the basic knowledge from probability theory and statistics. Probability En. Bayesian task and its two special cases, Bayesian task |

03 | 02.03.2022 | 09:15 | VH | Non-Bayesian tasks, formulations only. Conditional independence of features. Gaussian models. Non-bayesian tasks, Two statistical models. |

04 | 09.03.2022 | 09:15 | RŠ | Artificial neural networks. slides |

05 | 16.03.2022 | 09:15 | VH | Estimation of probabilistic models. Parametric and nonparametric methods. parametric, nonparametric |

06 | 23.03.2022 | 09:15 | RŠ | Convolutional neural networks. slides |

07 | 30.03.2022 | 09:15 | VH | Data normalization. Experimental evaluation of classifiers. Receiver operator curve (ROC). Experimental classifier performance evaluation |

08 | 06.04.2022 | 09:15 | VH | Learning in pattern recognition. VC dimension and its use. Learning in Pattern Recognition, four formulations, VC learning theory |

09 | 13.04.2022 | 09:15 | VH | Linear classifiers. Perceptron. Learning (training) algorithm. Linear classifiers. Perceptron. |

xx | 20.04.2022 | No lecture because two of four students could not come. | ||

10 | 27.04.2022 | 09:15 | VH | Support vector machines classifiers (SVM). Kernel methods. Training algorithms for SVM Linear SVM , Kernel SVMs |

11 | 4.05.2022 | 09:15 | VH | Unsupervised learning. K-means algorithm. EM algorithm. Unsupervised learning |

12 | 11.05.2022 | 09:15 | VH | Markovian models in pattern recognition. Markovian pattern recognition methods |

13 | 18.05.2022 | 09:15 | RŠ | Reinforcement learning. slides |

14 | 24.05.2022 | 09:15 | VH | Structural pattern recognition |

- The exam has two parts, a written test, and an oral part. The written test checks the students' overview of the subject. In a written part, the student answers typically five questions. The written test counts for 50 points maximally.
- In the oral part of the exam following the written part, the teacher finds a topic related to the student's research. paper. The aim of the oral exam is to check if the student knows some part of the subject topic in more detail. The oral part counts for 50 points maximally.
- The oral part of the exam follows the written part after the written part is corrected by the teacher.
- The resulting mark follows from the sum of points. The maximum number of points is 100. The evaluation Excellent is for 80 points and more. The evaluation Passed is for 51 points and more. Otherwise, the student failed the exam.

- Schlesinger M.I., Hlaváč V.: Ten lectures on statistical and structural pattern recognition, Kluwer Academic Publishers, Dordrecht, The Netherlands, 2002. Edition in Czech Vydavatelství ČVUT Praha, 1999.
- Duda R.O., Hart, P.E., Stork, D.G.: Pattern Classification, John Willey and sons, 2nd edition, New York, 2001.
- Bishop, C.M: Pattern Recognition and Machine Learning, Springer, New York, 2006.

courses/xp33rod/start.txt · Last modified: 2022/05/25 11:30 by hlavac