|
Lecturer(s)
|
-
Štursa Dominik, Ing. Ph.D.
-
Ksiažek Jakub, Ing.
-
Doležel Petr, prof. Ing. Ph.D.
|
|
Course content
|
Lecture topics by week of the semester: 1. Artificial neural networks - introduction, history, basic concepts. 2. Simple perceptron. 3. Hopfield network. 4. Kohohen's self-organizing map. 5. Multilayer perceptron I - definition of topology, error back propagation algorithm. 6. Multilayer perceptron II - Levenberg-Marquardt learning algorithm, application to approximation problems. 7. Multilayer perceptron III - use for modeling dynamic systems, use for process control. 8. Convolutional network - basic concepts and topology. 9. Convolutional network - classification. 10. Convolutional network - detection. 11. Convolutional network - instance and semantic segmentation. 12. Generative adversarial networks for image generation. 13. Use in industrial applications. Lecture topics by week of the semester: 1. Introduction to software tools. 2. Simple perceptron. 3. Hopfield network. 4. Kohohen's self-organizing map. 5. Multilayer perceptron I - topology implementation. 6. Multilayer perceptron II - implementation of the error back propagation algorithm. 7. Multilayer perceptron III - use for regression, approximation and prediction. 8. Convolutional network - basic concepts and topology. 9. Convolutional network - classification. 10. Convolutional network - detection. 11. Convolutional network - instance and semantic segmentation. 12. Generative adversarial networks for image generation. 13. Applications for quality control in manufacturing.
|
|
Learning activities and teaching methods
|
Monologic (reading, lecture, briefing), Methods of individual activities
- Contact teaching
- 52 hours per semester
- Preparation for an exam
- 17 hours per semester
- Home preparation for classes
- 26 hours per semester
- Individual project
- 55 hours per semester
|
|
Learning outcomes
|
The aim of the course to familiarize the students with the basic concepts of artificial neural networks and their applications.
Student will be able to create artificial neural networks and realize computationally their learning and implementation. Student will be able to apply stochastic optimization techniques to solve various tasks.
|
|
Prerequisites
|
There is expected fundamental knowledge of programming and graph theory.
|
|
Assessment methods and criteria
|
Oral examination, Home assignment evaluation
Obtaining credit is conditional on the preparation of a final project in which a set of problems from the discussed issues will be solved. The examination takes the form of an oral interview.
|
|
Recommended literature
|
-
Haykin, Simon S. Neural networks and learning machines. Upper Saddle River: Prentice Hall, 2009. ISBN 978-0-13-147139-9.
-
LAWLESS, William, Ranjeey MUTTU, Donald; SOFGE, Ira S. MISKOWITZ a Stephen RUSSELL. Artiticial Intellignece for the Internet of Everything. London: Elsevier, 2019. ISBN 978-0-1281-7636-8.
-
LUCCI, Stephen a Danny KOPEC. Artificial Intelligence in the 21st Centruy. 2nd Edition. Herndon: Mercury Learning and Information, 2016. ISBN 978-1-942270-00-3.
-
Nguyen, Hung T. A first course in fuzzy and neural control. Boca Raton: Chapman & Hall, 2003. ISBN 1-58488-244-1.
-
Olej, Vladimír. Úvod do umělé inteligence : moderní přístupy : distanční opora. Pardubice: Univerzita Pardubice, 2010. ISBN 978-80-7395-307-2.
-
Škrabánek, Pavel. Teorie fuzzy množin a jejich aplikace (online).. Pardubice: Univerzita Pardubice, 2014. ISBN 978-80-7395-875-6.
-
Zelinka, Ivan. Evoluční výpočetní techniky : principy a aplikace. Praha: BEN - technická literatura, 2009. ISBN 978-80-7300-218-3.
|