Lecturer(s)
|
-
Bukovský Ivo, doc. Ing. Ph.D.
-
Skrbek Miroslav, Ing. Ph.D.
-
Jelínek Jiří, Ing. CSc.
|
Course content
|
1. Introduction, overview of neural networks 2. Advanced feed-forward neural network architectures 3. Recurrent and LSTM neural networks 4. Neural Turing machines 5. Fuzzy systems and neural fuzzy systems 6. Reinforcement learning, actor-critic architecture 7. Evolutionary techniques and nature inspired optimization The topics solved in the labs follow the lecture topics. Students obtain their practical skills and experiences with advanced neural network architectures and nature-inspired optimization techniques in the labs. Students will benefit from commonly used neural frameworks, and they will pass through a series of laboratory tasks accomplished with a final project.
|
Learning activities and teaching methods
|
Monologic (reading, lecture, briefing), Work with text (with textbook, with book), Work with multi-media resources (texts, internet, IT technologies), Individual tutoring, Practical training, Blended learning
- Preparation for classes
- 26 hours per semester
- Semestral paper
- 25 hours per semester
- Preparation for credit
- 19 hours per semester
- Class attendance
- 30 hours per semester
|
Learning outcomes
|
This course is focused on advanced topic of computation intelligence algorithms. Students obtain practical experience with advanced recurrent neural networks, reinforcement learning, fuzzy systems and nature inspired optimization. The emphasis is on practical experience and individual student projects.
Upon successful completion of the course, students will gain knowledge of methods and tools of computational intelligence, demonstrate their ability of independent problem-solving, improve their skills of analytical thinking, algorithmization, use of machine learning tools, gain skills in design and SW implementation of neural networks, their training, and necessary work with data.
|
Prerequisites
|
Basics of programming (any programming language, Python and basic knowledge of neural networks an advantage), knowledge of mathematics of basic bachelor's courses.
|
Assessment methods and criteria
|
Test, Interim evaluation
To pass the course, students must meet at least 50% of the total evaluation of laboratory tasks and the semester project and achieve at least 50% of the overall evaluation of the final test. 26 hours: Home preparation 25 hours: Semester work 10 hours: Preparation for the exam Additional study materials: Materials for lectures and tutorials will be in USB LMS Moodle.
|
Recommended literature
|
-
CHARU C. AGGARWAL. Neural Networks and Deep Learning: A Textbook. Springer; 1st ed. 2018. ISBN: 978-3319944623. ISBN 978-3319944623.
-
LOTFI A ZADEH, RAFIK A ALIEV. Fuzzy Logic Theory and Applications: Part I and Part II. WSPC 2018. ISBN: 978-9813238176.
-
SIMEON KOSTADINOV. Recurrent Neural Networks with Python Quick Start Guide: Sequential learning and language modeling with TensorFlow. Packt Publishing, 2018. ISBN: 978-1789132335.
-
XIN-SHE YANG. Applied Reinforcement Learning with Python: With OpenAI Gym, Tensorflow, and Keras Paperback - Apress, 1st edition, 2019. ISBN: 978-1484251263.
-
XIN-SHE YANG. Nature-Inspired Optimization Algorithms, 2nd Edition. Academic Press, September 2020. 978-0128219867.
|