当前位置: X-MOL 学术Signal Image Video Process. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Joint distinct subspace learning and unsupervised transfer classification for visual domain adaptation
Signal, Image and Video Processing ( IF 2.3 ) Pub Date : 2020-07-29 , DOI: 10.1007/s11760-020-01745-w
Shiva Noori Saray , Jafar Tahmoresnezhad

In many real-world knowledge transfer and transfer learning scenarios, the known common problem is distribution discrepancy (i.e., the difference in type, distribution and dimensionality of features) between source and target domains. In this paper, we introduce joint distinct subspace learning and unsupervised transfer classification for visual domain adaptation (JDSC) method, which is an iterative two-step framework. JDSC is based on hybrid of feature-based and classifier-based approaches that uses the feature-based techniques to tackle the challenge of domain shift and classifier-based techniques to learn a reliable model. In addition, for subspace alignment, weighted joint geometrical and statistical alignment is proposed to learn two coupled projections for mapping the source and target data into respective subspaces by accounting the importance of marginal and conditional distributions, differently. The proposed method has been evaluated on various real-world image datasets. JDSC gets 86.2% average classification accuracy on four standard domain adaptation benchmarks. The experiments demonstrate that our proposed method achieves a significant improvement compared to other state of the arts in average classification accuracy. Our source code is available at https://github.com/jtahmores/JDSC .

中文翻译:

用于视觉域适应的联合不同子空间学习和无监督转移分类

在许多现实世界的知识迁移和迁移学习场景中,已知的常见问题是源域和目标域之间的分布差异(即特征的类型、分布和维度的差异)。在本文中,我们介绍了用于视觉域适应 (JDSC) 方法的联合不同子空间学习和无监督转移分类,这是一个迭代的两步框架。JDSC 基于基于特征和基于分类器的方法的混合,该方法使用基于特征的技术来解决域转移和基于分类器的技术来学习可靠模型的挑战。此外,对于子空间对齐,加权联合几何和统计对齐被提出来学习两个耦合投影,通过不同地考虑边缘和条件分布的重要性,将源和目标数据映射到各自的子空间。所提出的方法已在各种真实世界的图像数据集上进行了评估。JDSC 在四个标准域自适应基准上获得了 86.2% 的平均分类准确率。实验表明,与其他现有技术相比,我们提出的方法在平均分类精度方面取得了显着提高。我们的源代码可在 https://github.com/jtahmores/JDSC 获得。在四个标准域适应基准上平均分类准确率为 2%。实验表明,与其他现有技术相比,我们提出的方法在平均分类精度方面取得了显着提高。我们的源代码可在 https://github.com/jtahmores/JDSC 获得。在四个标准域适应基准上平均分类准确率为 2%。实验表明,与其他现有技术相比,我们提出的方法在平均分类精度方面取得了显着提高。我们的源代码可在 https://github.com/jtahmores/JDSC 获得。
更新日期:2020-07-29
down
wechat
bug