Knowledge Commons of Institute of Automation,CAS
Factorized Learning Assisted with Large Language Model for Gloss-free Sign Language Translation | |
Chen ZG(陈志刚)1,2; Zhou BJ(周本加)3; Li J(李俊)1,2; Wan J(万军)1,2,3; Lei Z(雷震)1,2,4; Jiang N(江宁)5; Lu Q(卢泉)5; Zhao GY(赵国营)5 | |
2024-05 | |
会议名称 | Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) |
会议日期 | 2024-5-22 |
会议地点 | Torino, Italia |
摘要 | Previous Sign Language Translation (SLT) methods achieve superior performance by relying on gloss annotations. However, labeling high-quality glosses is a labor-intensive task, which limits the further development of SLT. Although some approaches work towards gloss-free SLT through jointly training the visual encoder and translation network, these efforts still suffer from poor performance and inefficient use of the powerful Large Language Model (LLM). Most seriously, we find that directly introducing LLM into SLT will lead to insufficient learning of visual representations as LLM dominates the learning curve. To address these problems, we propose Factorized Learning assisted with Large Language Model (FLa-LLM) for gloss-free SLT. Concretely, we factorize the training process into two stages. In the visual initialing stage, we employ a lightweight translation model after the visual encoder to pre-train the visual encoder. In the LLM fine-tuning stage, we freeze the acquired knowledge in the visual encoder and integrate it with a pre-trained LLM to inspire the LLM’s translation potential. This factorized training strategy proves to be highly effective as evidenced by significant improvements achieved across three SLT datasets which are all conducted under the gloss-free setting. |
七大方向——子方向分类 | 生物特征识别 |
国重实验室规划方向分类 | 视觉信息处理 |
是否有论文关联数据集需要存交 | 否 |
文献类型 | 会议论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/56599 |
专题 | 多模态人工智能系统全国重点实验室_生物识别与安全技术 |
通讯作者 | Wan J(万军) |
作者单位 | 1.MAIS, Institute of Automation, Chinese Academy of Sciences 2.School of Artificial Intelligence, University of Chinese Academy of Sciences 3.Macau University of Science and Technology 4.CAIR, HKISI, Chinese Academy of Sciences, Hong Kong 5.Mashang Consumer Finance |
第一作者单位 | 中国科学院自动化研究所 |
通讯作者单位 | 中国科学院自动化研究所 |
推荐引用方式 GB/T 7714 | Chen ZG,Zhou BJ,Li J,et al. Factorized Learning Assisted with Large Language Model for Gloss-free Sign Language Translation[C],2024. |
条目包含的文件 | ||||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
Factorized Learning (800KB) | 会议论文 | 开放获取 | CC BY-NC-SA | 浏览 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论