Course detail

Machine Learning

FEKT-MPA-MLRAcad. year: 2021/2022

Students will gain insight into advanced machine learning methods. They will be able to describe and compare the properties of individual approaches to data classification. They will be able to select and apply a specific approach to a given problem. They will also gain practical experience with current implementations of machine learning methods including deep learning.

Language of instruction

English

Number of ECTS credits

5

Mode of study

Not applicable.

Guarantor

doc. Ing. Radim Kolář, Ph.D.

Department

Department of Biomedical Engineering (UBMI)

Offered to foreign students

The home faculty only

Learning outcomes of the course unit

Not applicable.

Prerequisites

Not applicable.

Co-requisites

Not applicable.

Planned learning activities and teaching methods

Not applicable.

Assesment methods and criteria linked to learning outcomes

A maximum of 30 points can be obtained during the semester. A maximum of 70 points can be obtained in the final examination.

There will be 6 tests during the semester (each test for maximum of 5 points). The tests cannot be repeated.

The conditions for the award of credit are as follows:
- full participation in the computer labs (max. two excused absences),
- obtaining at least 15 points from the tests.

Obtaining credit is a condition for admission to the final examination.

The final exam will be marked with a maximum of 70 points. A minimum of 35 points is required to pass the exam.

Course curriculum

1. Introduction to classification. Classification error, classifier testing.
2. Features assessment, feature selection and feature reduction with basic and advanced methods (PCA, mRMR, t-SNE).
3. Linear Classifiers - basic principles and methods (perceptron, SVM, MSE).
4. Kernel approach for non-linear classification.
5. Bayesian approach to classification. Naive Bayes classifier.
6. Maximum likelihood and Maximum a-posteriori probability.
7. Decision and regression trees and forests, random forests.
8. Methods for improving classifier properties (bagging, boosting).
9. Basics of neural networks, regularization.
10. Principles of deep learning, deep neural networks (NN) and basic building blocks.
11. Principles of deep NN learning.
12. Variants of deep NN, autoencoders, recurrent NN, LSTM, GRU, GAN.
13. Application of classification tasks for processing of signals, images and bioinformatic data. Application examples.

Work placements

Not applicable.

Aims

Not applicable.

Specification of controlled education, way of implementation and compensation for absences

Not applicable.

Recommended optional programme components

Not applicable.

Prerequisites and corequisites

Not applicable.

Basic literature

N. Buduma: Fundamentals of Deep Learning, O'Reilly Media, 2017 (CS)
Ch. M. Bishop: Pattern Recognition and Machine Learning, Springer, 2011
I. Goodfellow, Y. Bengio, A. Courville, F. Bach: Deep Learning, The MIT Press, 2016
Geron, A.: Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems, 2. edition, O'Reilly Media (EN)
Deisenroth,M.P, Faisal, A.A, Ong, Ch.S.:Mathematics for Machine Learning, Cambridge University Press, 2020 (EN)

Recommended reading

Not applicable.

eLearning

eLearning: currently opened course

Classification of course in study plans

  • Programme MPA-BIO Master's, 2. year of study, winter semester, compulsory
  • Programme MPC-BIO Master's, 2. year of study, winter semester, compulsory
  • Programme MPC-BTB Master's, 2. year of study, winter semester, compulsory

Type of course unit

 

Lecture

26 hours, optionally

Teacher / Lecturer

Ing. Jiří Chmelík, Ph.D.
doc. Ing. Radim Kolář, Ph.D.
Ing. Tomáš Vičar, Ph.D.

Exercise in computer lab

26 hours, compulsory

Teacher / Lecturer

Ing. Jakub Nemček
Ing. Tomáš Vičar, Ph.D.

eLearning

eLearning: currently opened course