arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1707.02892cs.CL
收藏

基于多任务学习的广义递归神经网络文本分类

A Generalized Recurrent Neural Architecture for Text Classification with Multi-Task Learning

Shanghai Jiao Tong University(上海交通大学)

Honglun Zhang, Liqiang Xiao, Yongkun Wang, Yaohui Jin

摘要:

多任务学习利用相关任务之间的潜在相关性来提取共同特征并产生性能增益。然而,大多数以前的作品只考虑简单或弱的相互作用,从而无法模拟三个或更多的任务之间的复杂的相关性。在本文中,我们提出了一种多任务学习架构,该架构具有四种类型的递归神经层,以融合多个相关任务之间的信息。该体系结构是灵活的,并考虑到各种任务之间的相互作用,这可以被视为许多以前的作品的一般情况下。在五个文本分类基准数据集上的大量实验表明,我们的模型可以显着提高相关任务的性能,并从其他人那里获得额外的信息。

英文摘要:

Multi-task learning leverages potential correlations among related tasks to extract common features and yield performance gains. However, most previous works only consider simple or weak interactions, thereby failing to model complex correlations among three or more tasks. In this paper, we propose a multi-task learning architecture with four types of recurrent neural layers to fuse information across multiple related tasks. The architecture is structurally flexible and considers various interactions among tasks, which can be regarded as a generalized case of many previous works. Extensive experiments on five benchmark datasets for text classification show that our model can significantly improve performances of related tasks with additional information from others.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑