Institutional Repository of Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
Structurally Comparative Hinge Loss for Dependency-Based Neural Text Representation | |
Wang, Kexin1,2; Zhou, Yu1,2,3; Zhang, Jiajun1,2; Wang, Shaonan1,2; Zong, Chengqing1,2 | |
发表期刊 | ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING |
ISSN | 2375-4699 |
2020-05 | |
期号 | 4页码:19 |
文章类型 | 期刊论文 |
摘要 | Dependency-based graph convolutional networks (DepGCNs) are proven helpful for text representation to handle many natural language tasks. Almost all previous models are trained with cross-entropy (CE) loss, which maximizes the posterior likelihood directly. However, the contribution of dependency structures is not well considered by CE loss. As a result, the performance improvement gained by using the structure information can be narrow due to the failure in learning to rely on this structure information. To face the challenge, we propose the novel structurally comparative hinge (SCH) loss function for DepGCNs. SCH loss aims at enlarging the margin gained by structural representations over non-structural ones. From the per- spective of information theory, this is equivalent to improving the conditional mutual information of model decision and structure information given text. Our experimental results on both English and Chinese datasets show that by substituting SCH loss for CE loss on various tasks, for both induced structures and structures from an external parser, performance is improved without additional learnable parameters. Furthermore, the extent to which certain types of examples rely on the dependency structure can be measured directly by the learned margin, which results in better interpretability. In addition, through detailed analysis, we show that this structure margin has a positive correlation with task performance and structure induction of DepGCNs, and SCH loss can help model focus more on the shortest dependency path between entities. We achieve the new state-of-the-art results on TACRED, IMDB, and Zh. Literature datasets, even compared with ensemble and BERT baselines. |
关键词 | Text representation graph convolutional networks loss function |
DOI | 10.1145/3387633 |
收录类别 | SCI |
语种 | 英语 |
WOS记录号 | WOS:000582616900012 |
七大方向——子方向分类 | 自然语言处理 |
引用统计 | |
文献类型 | 期刊论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/39115 |
专题 | 模式识别国家重点实验室_自然语言处理 |
通讯作者 | Wang, Kexin |
作者单位 | 1.National Laboratory of Pattern Recognition, Institute of Automation, CAS 2.University of Chinese Academy of Sciences, Beijing 100049, P. R. China 3.Beijing Fanyu Technology Co., Ltd |
第一作者单位 | 模式识别国家重点实验室 |
通讯作者单位 | 模式识别国家重点实验室 |
推荐引用方式 GB/T 7714 | Wang, Kexin,Zhou, Yu,Zhang, Jiajun,et al. Structurally Comparative Hinge Loss for Dependency-Based Neural Text Representation[J]. ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING,2020(4):19. |
APA | Wang, Kexin,Zhou, Yu,Zhang, Jiajun,Wang, Shaonan,&Zong, Chengqing.(2020).Structurally Comparative Hinge Loss for Dependency-Based Neural Text Representation.ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING(4),19. |
MLA | Wang, Kexin,et al."Structurally Comparative Hinge Loss for Dependency-Based Neural Text Representation".ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING .4(2020):19. |
条目包含的文件 | 下载所有文件 | |||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
Structurally Compara(1553KB) | 期刊论文 | 作者接受稿 | 开放获取 | CC BY-NC-SA | 浏览 下载 |
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