Institutional Repository of Vilnius University
Gilieji neuroniniai tinklai ir jų taikymas /
Abstract
dc:descriptionIn the theoretical part of the bachelor’s work, neural networks are analyzed, their learning types, how they are trained and tested, and the indicators that determine the accuracy of the model are named. The structure of a deep neural network and when artificial intelligence can be called deep artificial intelligence are also described. Also, one of the main topics analyzed in this work is convolutional neural networks, its structure and the principle of operation of layers. The activation functions of the layers are described and thus how to improve the performance of the neural network when dealing with overfitting problems is described. Also, the paper analyzes transfer learning, its advantages and describes one of the best types of model architectures in transfer learning. In the researched part of the bachelor’s degree, a study was carried out – data classification. In the course of this work, image classification was performed using transfer learning using the Ef- ficientNet model architecture with the noisy-student method. During the research, a model capable of classification of apple tree leaf photos was developed with high precission.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Rutkovskij, Jaroslav,
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:ELABAETD192845431&prefLang=en_US