Deterministic Reversible Data Augmentation for Neural Machine Translation
J Yao, H Huang, Z Liu, Y Guo - arXiv preprint arXiv:2406.02517, 2024 - arxiv.org
Data augmentation is an effective way to diversify corpora in machine translation, but
previous methods may introduce semantic inconsistency between original and augmented …
previous methods may introduce semantic inconsistency between original and augmented …
DiverSeg: Leveraging Diverse Segmentations with Cross-granularity Alignment for Neural Machine Translation
In this study, we proposed DiverSeg to exploit diverse segmentations from multiple subword
segmenters that capture the various perspectives of each word for neural machine …
segmenters that capture the various perspectives of each word for neural machine …
SubMerge: Merging Equivalent Subword Tokenizations for Subword Regularized Models in Neural Machine Translation
Subword regularized models leverage multiple subword tokenizations of one target
sentence during training. Previous decoding algorithms select one tokenization during …
sentence during training. Previous decoding algorithms select one tokenization during …
[PDF][PDF] Studies on Subword-based Low-Resource Neural Machine Translation: Segmentation, Encoding, and Decoding
S Haiyue - 2024 - repository.kulib.kyoto-u.ac.jp
In a world rich with diverse ideas and cultures, humans are isolated into islands of distinct
languages. Machine translation (MT) serves as a bridge, facilitating information access and …
languages. Machine translation (MT) serves as a bridge, facilitating information access and …