arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2012.06236cs.CL
收藏

利用常识知识改进零样本学习基线

Improving Zero Shot Learning Baselines with Commonsense Knowledge

Nanyang Technological University(南洋理工大学);Singapore University of Technology and Design(新加坡科技设计大学);University of Michigan(密歇根大学)

Abhinaba Roy, Deepanway Ghosal, Erik Cambria, Navonil Majumder, Rada Mihalcea, Soujanya Poria

摘要:

零样本学习--在完全不相交的类集合上进行训练和测试的问题--在很大程度上依赖于它将知识从训练类转移到测试类的能力。传统的语义嵌入由人类定义的属性(HA)或分布式词嵌入(DWE),用于促进这种转移,通过改善视觉和语义嵌入之间的关联。在本文中,我们利用显式的关系定义在ConceptNet,常识知识图谱,生成常识嵌入的类标签,通过使用基于图卷积网络的自动编码器。当我们将常识嵌入与现有的语义嵌入(即HA和DWE)融合时,我们在三个标准基准数据集上进行的实验超过了强基线。

英文摘要:

Zero shot learning -- the problem of training and testing on a completely disjoint set of classes -- relies greatly on its ability to transfer knowledge from train classes to test classes. Traditionally semantic embeddings consisting of human defined attributes (HA) or distributed word embeddings (DWE) are used to facilitate this transfer by improving the association between visual and semantic embeddings. In this paper, we take advantage of explicit relations between nodes defined in ConceptNet, a commonsense knowledge graph, to generate commonsense embeddings of the class labels by using a graph convolution network-based autoencoder. Our experiments performed on three standard benchmark datasets surpass the strong baselines when we fuse our commonsense embeddings with existing semantic embeddings i.e. HA and DWE.

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。

↑