SYLLABUS
University: Technical University of Košice
Faculty: Faculty of Electrical Engineering and Informatics
Department: Department of Computers and Informatics
Course Number: 26000887 Course Name: Stochastic modeling and data analysis
Type, scope and method of learning activities:
Course Type: Lecture, Laboratory exercise
Recommended scope of the course content (in hours):
Full-time study (hours per week): 2,2
Part-time study (hours per semester): 26,26
Study Method:
Number of credits: 6
Recommended semester of study: ST
Recommended semester Study programme Study grade Study Method
1.rok ST Cybersecurity (KB_Ing_D_sk)
Informatics (INF_Ing_D_sk)
Informatics (INF_Ing_D_en)
Master
Master
Master
Attendance
Attendance
Attendance
Level of study:
Prerequisites:
Course completion requirements:
Assessment and completion of the course: Credit test and examination
Continuous assessment: Student passes the continuous assessment and receives credits when he or she meets the requirement to obtain at least 21% out of 40%.
Credit test
Final assessment: Student passes the final assessment and passes the examination when he or she meets the requirement to obtain at least 31% out of 60%.
Examination
Overall assessment: Overall assessment is the sum of the assessments obtained by students in the assessment period. The overall result is determined in accordance with the internal regulations of the Technical University in Košice. (Study Regulations, the internal regulation principles of doctoral studies)
Learning outcomes:
The subject aims to acquire practical knowledge for solving complex problems using machine learning methods. The student will become familiar with more advanced principles in machine learning and artificial intelligence, emphasizing current trends in this field. The student will acquire the knowledge necessary for applications such as image recognition or sentiment analysis from the text. The student will also gain essential skills for learning agent bots in dynamic environments (computer games, OpenAI environment) using primary forms of machine learning by reinforcement (Q and SARSA learning), as well as their advanced designs using deep neural networks (DQN, double DQN, Actor-critic and other).
Brief course content:
1. Neural networks and deep learning.
 2. Multi-layer neural networks with forward propagation.
 3. Error backpropagation method, activation functions.
 4. Convolutional neural networks, image recognition.
 5. Recurrent neural networks.
 6. Autoencoder.
 7. Introduction to the issue of dynamic systems and a machine learning perspective on systems defined in this way.
8. Bellman equations and Markov systems.
9. Simple forms of machine learning in dynamic systems and the OpenAI environment.
10. Machine learning with reinforcement, Q and SARSA learning.
11. Deep learning with reinforcement using neural networks.
12. Other forms of learning with reinforcement using advanced methods (Actor-critic and DDPG approach).
Recommended Reference Sources:
Python for Data Analysis, Wes McKinney, O'Reilly Media, 2012
Machine Learning: An algorithmic perspective, Stephen Marsland, CRC Press, 2015
Agent Based and Individual based modelling, Steven F. Railsback and Volker Grimm, Princeton Press, 2012
Artificial Intelligence for Humans, Jeff Heaton, Heaton Research, 2014  
Recommended optional program components:
Languages required for the course completion:
Notes:
Course assessment:
Total number of students assessed: 230
  A B C D E FX  
  19% 17% 18% 19% 12% 16%  
Teacher:
prof. Ing. Peter Drotár, PhD.
prof. Ing. Juraj Gazda, PhD.
Last modified: 01.09.2022
Approved by: person(s) responsible for the study program