| 基于伪成对标签的深度无监督哈希学习 |
| |
| 引用本文： | 林计文,刘华文. 基于伪成对标签的深度无监督哈希学习[J]. 模式识别与人工智能, 2020, 33(3): 258-267. DOI: 10.16451/j.cnki.issn1003-6059.202003007 |
| |
| 作者姓名： | 林计文 刘华文 |
| |
| 作者单位： | 1.浙江师范大学 数学与计算机科学学院 金华 321004 |
| |
| 基金项目： | 国家自然科学基金项目(No.61572443);浙江省自然科学基金项目(No.LY14F020019)资助。 |
| |
| 摘 要： | 无监督的深度哈希学习方法由于缺少相似性监督信息,难以获取高质量的哈希编码.因此,文中提出端到端的基于伪成对标签的深度无监督哈希学习模型.首先对由预训练的深度卷积神经网络得到的图像特征进行统计分析,用于构造数据的语义相似性标签.再进行基于成对标签的有监督哈希学习.在两个常用的图像数据集CIFAR-10、NUS-WIDE上的实验表明,经文中方法得到的哈希编码在图像检索上的性能较优.
|
| 关 键 词： | 哈希学习 深度无监督哈希学习 伪标签 近似最近邻搜索 图像检索 |
| 收稿时间： | 2019-11-12 |
Deep Unsupervised Hashing with Pseudo Pairwise Labels |
| |
| LIN Jiwen,LIU Huawen. Deep Unsupervised Hashing with Pseudo Pairwise Labels[J]. Pattern Recognition and Artificial Intelligence, 2020, 33(3): 258-267. DOI: 10.16451/j.cnki.issn1003-6059.202003007 |
| |
| Authors: | LIN Jiwen LIU Huawen |
| |
| Affiliation: | 1.College of Mathematics and Computer Science, Zhejiang Normal University, Jinhua 321004 |
| |
| Abstract: | It is difficult to obtain high-quality hash codes for unsupervised deep hashing methods due to the lack of similarity supervised information.Therefore,an end-to-end deep unsupervised hashing model based on pseudo-pairwise labels is proposed.Statistical analysis is performed on the image features extracted by the pre-trained deep convolutional neural network to construct the semantic similarity labels for data.Supervised deep hashing based on pairwise labels is then conducted.Experiments on commonly used image datasets CIFAR-10 and NUS-WIDE indicate that hash codes obtained by the proposed method perform better on image retrieval. |
| |
| Keywords: | Learning to Hash Deep Unsupervised Hashing Pseudo Label Approximate Nearest Neighbor Search Image Retrieval |
| 本文献已被 维普 万方数据 等数据库收录！ |
| 点击此处可从《模式识别与人工智能》浏览原始摘要信息 |
|
点击此处可从《模式识别与人工智能》下载全文 |
