A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation
In a real-world application, the images taken by different cameras with different conditions often incur illumination variation, low-resolution, different poses, blur, etc., which leads to a large distribution difference or gap between training (source) and test (target) images. This distribution ga...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-08-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/20/16/4367 |
| _version_ | 1827711522272444416 |
|---|---|
| author | Rakesh Kumar Sanodiya Leehter Yao |
| author_facet | Rakesh Kumar Sanodiya Leehter Yao |
| author_sort | Rakesh Kumar Sanodiya |
| collection | DOAJ |
| description | In a real-world application, the images taken by different cameras with different conditions often incur illumination variation, low-resolution, different poses, blur, etc., which leads to a large distribution difference or gap between training (source) and test (target) images. This distribution gap is challenging for many primitive machine learning classification and clustering algorithms such as k-Nearest Neighbor (k-NN) and k-means. In order to minimize this distribution gap, we propose a novel Subspace based Transfer Joint Matching with Laplacian Regularization (STJML) method for visual domain adaptation by jointly matching the features and re-weighting the instances across different domains. Specifically, the proposed STJML-based method includes four key components: (1) considering subspaces of both domains; (2) instance re-weighting; (3) it simultaneously reduces the domain shift in both marginal distribution and conditional distribution between the source domain and the target domain; (4) preserving the original similarity of data points by using Laplacian regularization. Experiments on three popular real-world domain adaptation problem datasets demonstrate a significant performance improvement of our proposed method over published state-of-the-art primitive and domain adaptation methods. |
| first_indexed | 2024-03-10T17:56:47Z |
| format | Article |
| id | doaj.art-36cbcc1184934dce9441b1feee8f38ec |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T17:56:47Z |
| publishDate | 2020-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-36cbcc1184934dce9441b1feee8f38ec2023-11-20T09:09:19ZengMDPI AGSensors1424-82202020-08-012016436710.3390/s20164367A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain AdaptationRakesh Kumar Sanodiya0Leehter Yao1Department of Electrical Engineering, National Taipei University of Technology, Taipei 10608, TaiwanDepartment of Electrical Engineering, National Taipei University of Technology, Taipei 10608, TaiwanIn a real-world application, the images taken by different cameras with different conditions often incur illumination variation, low-resolution, different poses, blur, etc., which leads to a large distribution difference or gap between training (source) and test (target) images. This distribution gap is challenging for many primitive machine learning classification and clustering algorithms such as k-Nearest Neighbor (k-NN) and k-means. In order to minimize this distribution gap, we propose a novel Subspace based Transfer Joint Matching with Laplacian Regularization (STJML) method for visual domain adaptation by jointly matching the features and re-weighting the instances across different domains. Specifically, the proposed STJML-based method includes four key components: (1) considering subspaces of both domains; (2) instance re-weighting; (3) it simultaneously reduces the domain shift in both marginal distribution and conditional distribution between the source domain and the target domain; (4) preserving the original similarity of data points by using Laplacian regularization. Experiments on three popular real-world domain adaptation problem datasets demonstrate a significant performance improvement of our proposed method over published state-of-the-art primitive and domain adaptation methods.https://www.mdpi.com/1424-8220/20/16/4367domain adaptationunsupervised discriminant analysistransfer learningclassificationfeature learninginstance re-weighting |
| spellingShingle | Rakesh Kumar Sanodiya Leehter Yao A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation Sensors domain adaptation unsupervised discriminant analysis transfer learning classification feature learning instance re-weighting |
| title | A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation |
| title_full | A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation |
| title_fullStr | A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation |
| title_full_unstemmed | A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation |
| title_short | A Subspace Based Transfer Joint Matching with Laplacian Regularization for Visual Domain Adaptation |
| title_sort | subspace based transfer joint matching with laplacian regularization for visual domain adaptation |
| topic | domain adaptation unsupervised discriminant analysis transfer learning classification feature learning instance re-weighting |
| url | https://www.mdpi.com/1424-8220/20/16/4367 |
| work_keys_str_mv | AT rakeshkumarsanodiya asubspacebasedtransferjointmatchingwithlaplacianregularizationforvisualdomainadaptation AT leehteryao asubspacebasedtransferjointmatchingwithlaplacianregularizationforvisualdomainadaptation AT rakeshkumarsanodiya subspacebasedtransferjointmatchingwithlaplacianregularizationforvisualdomainadaptation AT leehteryao subspacebasedtransferjointmatchingwithlaplacianregularizationforvisualdomainadaptation |