This paper investigates transfer learning methods for low-resource Luo languages, focusing on the similarity of the languages. The study trains a more resourced language pair (parent model) and transfers some learned parameters to the less-resourced pair (child model) to initialize and constrain training.
Metadata
Author
Ogayo, Atieno Peres
Supervisor
Clovis Wanziguya
Faculty
COMPUTER SCIENCE
Department
Not listed
Document type
Final Year Project
Year
2021
Keywords
Transfer learning, Neural Machine Translation, Machine Translation, Cross-lingual Transfer, transfer learning related, learning related low, related low neural, low neural machine, machine translation luo, translation luo languages
Access
View Only
License
all_rights_reserved
Cite this work
Auto-generated from the record metadata: authors, journal title, DOI, volume, issue/serial, pages and publication date. Review before final academic use.