450-6023/02 – Machine Learning Methods for Embedded Systems (MSUVS)
| Gurantor department | Department of Cybernetics and Biomedical Engineering | Credits | 10 |
| Subject guarantor | prof. Ing. Michal Prauzek, Ph.D. | Subject version guarantor | prof. Ing. Michal Prauzek, Ph.D. |
| Study level | postgraduate | Requirement | Choice-compulsory type B |
| Year | | Semester | winter + summer |
| | Study language | English |
| Year of introduction | 2021/2022 | Year of cancellation | |
| Intended for the faculties | HGF, FEI | Intended for study types | Doctoral |
Subject aims expressed by acquired skills and competences
Students will be able to implement machine learning methods for the needs of embedded systems applications focused on microcontrollers, microprocessors or programable gate arrays after completing the course. Furthermore, students will be able to simulate the selected method and estimate its behavior within the target application. In the field of optimization approaches, students will be able to apply optimization methods and use this approach to optimize an implementation itself or hardware resources.
Teaching methods
Lectures
Individual consultations
Project work
Summary
The aim of the course Machine learning methods for embedded systems is to introduce students current modern approaches in the field of machine learning methods that are directly implementable in computationally limited embedded systems. Students are also introduced to the possibilities of optimizing the development or function of embedded systems using machine learning methods. Machine learning methods are being used today in many application areas of embedded systems and microcontrollers, and this situation will continue in the future. In this course, students will learn practical applications and application of these modern approaches in real implementation.
Compulsory literature:
[1] Burkov, Andriy. Machine Learning Engineering. 2020.
[2] Burkov, Andriy. The Hundred-Page Machine Learning Book. 2019.
[3] Mitchell, Tom M. Machine Learning. McGraw Hill, 2017.
[4] Harrington, Peter. Machine learning in action. Manning Publications Co, 2012.
Recommended literature:
[1] Embedded deep learning: algorithms, architectures and circuits for always-on neural network processing. Springer Science+Business Media, 2018.
[2] Reinforcement learning. Springer Science+Business Media, 2017.
[3] Eiben, A. E., a J. E. Smith. Introduction to Evolutionary Computing. Springer Verlag, 2015.
[4] Aggarwal, Charu C., a Chandan K. Reddy, editoři. Data clustering: algorithms and applications. Chapman and Hall/CRC, 2014.
Additional study materials
Way of continuous check of knowledge in the course of semester
exam
E-learning
Other requirements
There are not defined other requirements for student.
Prerequisities
Subject has no prerequisities.
Co-requisities
Subject has no co-requisities.
Subject syllabus:
The course deals with areas of applications in embedded systems with implemented methods:
A) Unsupervised learning (Cluster analysis …)
B) Supervised learning (Support vector machines, regression linearization, decision trees, k-nearest neighbor algorithm, neural networks …)
C) Semi-supervised learning - methods of reinforcement learning (Markov's decision-making process, Q-learning …)
In terms of progressive optimization of the function and design of embedded systems, the cource inludes bio-inspired methods from a family of evolutionary algorithms (genetic algorithms, differential evolution, etc.) or other approaches such as particle swarm optimization.
The application areas of the course are focused to deployment in the field of Internet-of-Things devices, ultra-low power devices with energy harvesting from various domain such as sensor systems, applications associated with the concept of Industry 4.0, SmartCities and SmartMetering.
Conditions for subject completion
Occurrence in study plans
Occurrence in special blocks
Assessment of instruction
Předmět neobsahuje žádné hodnocení.