Natural Language Inference over Interaction Space
Yichen Gong‡†{{{}^{\dagger}}{{}^{\ddagger}}}, Heng Luo‡, Jian Zhang‡ † New York University, New York, USA ‡ Horizon Robotics, Inc., Beijing, China {heng.luo,
Abstract
Natural Language Inference (NLI) task requires an agent to determine the logical relationship between a natural language premise and a natural language hypothesis. We introduce Interactive Inference Network (IIN), a novel class of neural network architectures that is able to achieve high-level understanding of the sentence pair by hierarchically extracting semantic features from interaction space. We show that an interaction tensor (attention weight) contains semantic information to solve natural language inference, and a denser interaction tensor contains richer semantic information. One instance of such architecture, Densely Interactive Inference Network (DIIN), demonstrates the state-of-the-art performance on large scale NLI copora and large-scale NLI alike corpus. It’s noteworthy that DIIN achieve a greater than 20% error reduction on the challenging Multi-Genre NLI (MultiNLI; Williams et al. 2017) dataset with respect to the strongest published system.
中文速览
自然语言推断（Natural Language Inference, NLI）要求模型判断两句话之间是蕴含、矛盾还是中立关系，这既需要理解词汇语义，也需要掌握常识，一直是自然语言理解中的核心难题。研究者提出了一种名为"交互推断网络"（Interactive Inference Network, IIN）的新型神经网络框架，其核心思路是：把前提句和假设句的词向量做逐维度的外积，构建一个高阶"交互张量"（interaction tensor），再用类似图像识别中的DenseNet对这个张量进行层级卷积特征提取，从而在"交互空间"而非传统的句子表示空间里完成推断。框架的一个具体实例——密集交互推断网络（Densely Interactive Inference Network, DIIN）——在大规模标准数据集SNLI和MultiNLI上均刷新了最优成绩，其中在MultiNLI上相较此前最强系统的错误率降低了逾20%，在问题对匹配数据集Quora Question Pairs上也达到了新的最优水平。这项工作表明注意力权重本身就蕴含丰富的语义信息，将交互张量与强大的卷积特征提取器结合是解决复杂语义推断任务的有效新路径。
原文 arXiv:1709.04348;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1709.04348v2