Institutional Repository of Vilnius University
Neuroniniai tinklai mikrokontroleriuose /
Abstract
dc:descriptionThis paper examines Feed-forward multi-layer perception neural network with various data types neurons weights in microcontrollers. To achieve this goal, a number of different microcontrollers were introduced and compared by their types and properties. Also, a neural network with different data types, such as Double, Float, Float16 and Integer was created and runed in different microcontrollers. Various neural network activation functions were examined and compared by their performance in speed (network cycles per second). Investigated network size selection methodologies and recommendations. A network speed (network cycles per second) and size (maximum size network that fits) studies were conducted with each microcontroller. Then their performance was compared to find suitable Feed-forward neural network for each type of microcontroller.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Velevko, Ernestas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD146237396&prefLang=en_US