Institutional Repository of Vilnius University
Creation and evaluation of a browser-based federated learning system /
Abstract
dc:descriptionThis research focuses on the analysis, development and evaluation of browser-based federated learning system, with the emphasis on providing potential foundation for future applications of decentralised model training concepts. The main contribution is a website, that is made publicly available, for everyone to participate in decentralised training of handwritten character recogniser model. Users can contribute their written samples and device computational power for global model training and improvement, without exposing private data. By performing appropriate Federated Aggregation technique, server ensures that all updates even from most distinct clients will affect the model learning. The usage of Federated approach reduces computational needs for both clients and server, comparing with the centralized style, making model training and training participation more accessible and much more environmentally sustainable.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Arbatov, Daniil,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD277692650&prefLang=en_US