| 大会名称 |
| 2023年 総合大会 |
| 大会コ－ド |
| 2023G |
| 開催年 |
|
2023
|
| 発行日 |
|
2023-02-28
|
| セッション番号 |
|
D-20
|
| セッション名 |
|
情報論的学習理論と機械学習
|
| 講演日 |
|
2023/3/8
|
| 講演場所(会議室等) |
|
2号館 2401教室
|
| 講演番号 |
|
D-20-19
|
| タイトル |
|
Domain Adaptation for Japanese Sentence Embedding Models with Contrastive Learning
|
| 著者名 |
|
◎Zihao Chen, Hisashi Handa, Kimiaki Shirahama,
|
| キーワード |
|
Domain adaptation, Contrastive learning, Data generation, Low-resource language
|
| 抄録 |
|
We propose a novel Japanese sentence representation framework JCSE that creates training data for domain adaptation by generating sentences and synthesizing them with sentences available in a target domain. Specifically, a pre-trained data generator is finetuned to the target domain using our collected corpus. It is then used to generate contradictory sentence pairs that are used in contrastive learning for adapting a Japanese language model to a specific task in the target domain. The experimental results show that JCSE achieves significant performance improvement surpassing direct transfer and other training strategies. We believe that JCSE paves a practicable way for domain-specific downstream tasks in low-resource languages like Japanese.
|
| 本文pdf |
|
PDF download
|
