Neural Proto-Language Reconstruction

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Cui, Chenxuan, Chen, Ying, Wang, Qinxin, Mortensen, David R.
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911851676172288
author Cui, Chenxuan
Chen, Ying
Wang, Qinxin
Mortensen, David R.
author_facet Cui, Chenxuan
Chen, Ying
Wang, Qinxin
Mortensen, David R.
contents Proto-form reconstruction has been a painstaking process for linguists. Recently, computational models such as RNN and Transformers have been proposed to automate this process. We take three different approaches to improve upon previous methods, including data augmentation to recover missing reflexes, adding a VAE structure to the Transformer model for proto-to-language prediction, and using a neural machine translation model for the reconstruction task. We find that with the additional VAE structure, the Transformer model has a better performance on the WikiHan dataset, and the data augmentation step stabilizes the training.
format Preprint
id arxiv_https___arxiv_org_abs_2404_15690
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Neural Proto-Language Reconstruction
Cui, Chenxuan
Chen, Ying
Wang, Qinxin
Mortensen, David R.
Computation and Language
Machine Learning
Proto-form reconstruction has been a painstaking process for linguists. Recently, computational models such as RNN and Transformers have been proposed to automate this process. We take three different approaches to improve upon previous methods, including data augmentation to recover missing reflexes, adding a VAE structure to the Transformer model for proto-to-language prediction, and using a neural machine translation model for the reconstruction task. We find that with the additional VAE structure, the Transformer model has a better performance on the WikiHan dataset, and the data augmentation step stabilizes the training.
title Neural Proto-Language Reconstruction
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2404.15690