ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2020, Vol. 57 ›› Issue (8): 1707-1714.doi: 10.7544/issn1000-1239.2020.20200122

所属专题: 2020数据挖掘与知识发现专题

• 人工智能 • 上一篇    下一篇

基于双向线性回归的监督离散跨模态散列方法

刘兴波1,聂秀山2,尹义龙1   

  1. 1(山东大学软件学院 济南 250101);2(山东建筑大学计算机科学与技术学院 济南 250101) (sclxb@mail.sdu.edu.cn)
  • 出版日期: 2020-08-01
  • 基金资助: 
    国家自然科学基金项目(61671274,61876098,61701280,61573219);国家重点研发计划项目(2018YFC0830100,2018YFC0830102);中国博士后科学基金项目(2016M592190);山东建筑大学杰出教授特别基金项目

Mutual Linear Regression Based Supervised Discrete Cross-Modal Hashing

Liu Xingbo1, Nie Xiushan2, Yin Yilong1   

  1. 1(School of Software, Shandong University, Jinan 250101);2(School of Computer Science and Technology, Shandong Jianzhu University, Jinan 250101)
  • Online: 2020-08-01
  • Supported by: 
    This work was supported by the National Natural Science Foundation of China (61671274, 61876098, 61701280, 61573219), the National Key Research and Developinent Program of China (2018YFC0830100, 2018YFC0830102), the China Postdoctoral Science Foundation (2016M592190), and the Special Funds for Distinguished Professors of Shandong Jianzhu University.

摘要: 跨模态散列可以将异构的多模态数据映射为语义相似度保持的紧凑二值码,为跨模态检索提供了极大的便利.现有的跨模态散列方法在利用类别标签时,通常使用2个不同的映射来表示散列码和类别标签之间的关系.为更好地捕捉散列码和语义标签之间的关系,提出一种基于双向线性回归的监督离散型跨模态散列方法.该方法仅使用一个稳定的映射矩阵来描述散列码与相应标签之间线性回归关系,提升了跨模态散列学习精度和稳定性.此外,该方法在学习用于生成新样本散列码的模态特定映射时,充分考虑了异构模态的特征分布与语义相似度的保持.在2个公开数据集上与现有方法的实验结果验证了该方法在各种跨模态检索场景下的优越性.

关键词: 近邻检索, 跨模态检索, 散列学习, 有监督散列, 双向映射

Abstract: Cross-modal hashing can map heterogeneous multimodal data into compact binary codes with similarity preserving, which provides great efficiency in cross-modal retrieval. Existing cross-modal hashing methods usually utilize two different projections to describe the correlation between Hash codes and class labels. In order to capture the relation between Hash codes and semantic labels efficiently, we propose a method named mutual linear regression based supervised discrete cross-modal hashing(SDCH) in this study. Only one stable projection is used in the proposed method to describe the linear regression relation between Hash codes and corresponding labels, which enhances the precision and stability in cross-modal hashing. In addition, we learn the modality-specific projections for out-of-sample extension by preserving the similarity and considering the feature distribution with different modalities. Comparisons with several state-of-the-art methods on two benchmark datasets verify the superiority of SDCH under various cross-modal retrieval scenarios.

Key words: approximate nearest neighbour search, cross-modal retrieval, learning-based hashing, supervised hashing, mutual linear regression

中图分类号: