|
Lecturer(s)
|
-
Doležel Petr, prof. Ing. Ph.D.
|
|
Course content
|
Lecture topics by week of the semester: 1. Artificial neural networks, division, definition of terms. 2. A simple perceptron as an elementary neural network. 3. Multilayer perceptron I - definition, back propagation algorithm. 4. Multilayer perceptron II - universal approximation, modeling of dynamical systems, process control. 5. Introduction to image data processing. Process of digital image creation. Digital image representation, basic image editing operations. 6. Convolutional neural network I - signal and image processing. 7. Convolutional neural network II - industrial applications. 8. Fuzzy sets theory. Fuzzy relations and operations with fuzzy relations. Linguistic variable, fuzzy logic. 9. Fuzzy logic systems and their use for process control. 10. Creation of digital twins of technological processes. 11. Introduction to reinforcement learning. Fundamentals of grasping and manipulation. 12. Learning-based grasping and manipulation. 13. Case studies of industrial applications. The contents of the seminars correspond to the contents of the lectures.
|
|
Learning activities and teaching methods
|
Monologic (reading, lecture, briefing), Work with text (with textbook, with book), Skills training
- Home preparation for classes
- 47 hours per semester
- Individual project
- 41 hours per semester
- Contact teaching
- 12 hours per semester
- Preparation for an exam
- 20 hours per semester
|
|
Learning outcomes
|
The aim of the course is to introduce to student modern methods with basic paradigms of artificial neural networks and fuzzy theory along with their practical implementation for automation and process control.
Upon completion of the course, the student demonstrates knowledge and skills in modern methods of process control based on machine learning, deep learning and fuzzy inference. The student can apply this knowledge to solve other engineering problems.
|
|
Prerequisites
|
unspecified
|
|
Assessment methods and criteria
|
Oral examination, Self project defence
Earning credit is contingent on completing a semester project. The examination is conducted in the form of an oral interview.
|
|
Recommended literature
|
-
Neural networks and soft computing. Heidelberg: Physica-Verlag, 2003. ISBN 3-7908-0005-8.
-
AGGARWAL, Charu C. Neural networks and deep learning: a textbook. Cham, 2018. ISBN 978-3-319-94462-3.
-
CASTILLO, Oscar, Janusz KACPRZYK, William MELEK, Patricia MELIN a Marek REFORMAT. Fuzzy Logic in Intelligent System Design: Theory and Applications. 2018.
-
SUTTON, Richard S. a Andrew G. BARTO. Reinforcement learning: an introduction. Second edition. Cambridge: Massachusetts, 2018. ISBN 978-026-2039-246.
|