arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1602.04341cs.CL
收藏

基于注意力的卷积神经网络机器理解

Attention-Based Convolutional Neural Network for Machine Comprehension

University of Munich(慕尼黑大学)

Wenpeng Yin, Sebastian Ebert, Hinrich Schütze

摘要:

理解开放域文本是自然语言处理(NLP)面临的主要挑战之一。机器理解基准仅基于文本内容来评估系统理解文本的能力。在这项工作中,我们调查机器理解MCTest,一个问答(QA)基准。以前的工作主要是基于特征工程方法。我们提出了一个神经网络框架,命名为分层注意力为基础的卷积神经网络(HABCNN),以解决这个任务,没有任何手动设计的功能。具体来说,我们通过两条路线探索HABCNN,一条是通过传统的通道,问题和答案的联合建模,一条是通过文本蕴涵。HABCNN采用注意力机制来检测与回答问题相关的关键短语、关键句子和关键片段。实验表明,HABCNN的性能大大优于之前的深度学习方法。

英文摘要:

Understanding open-domain text is one of the primary challenges in natural language processing (NLP). Machine comprehension benchmarks evaluate the system's ability to understand text based on the text content only. In this work, we investigate machine comprehension on MCTest, a question answering (QA) benchmark. Prior work is mainly based on feature engineering approaches. We come up with a neural network framework, named hierarchical attention-based convolutional neural network (HABCNN), to address this task without any manually designed features. Specifically, we explore HABCNN for this task by two routes, one is through traditional joint modeling of passage, question and answer, one is through textual entailment. HABCNN employs an attention mechanism to detect key phrases, key sentences and key snippets that are relevant to answering the question. Experiments show that HABCNN outperforms prior deep learning approaches by a big margin.

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。

↑