Towards Making High Dimensional Distance Metric Learning Practical
In this work, we study distance metric learning (DML) for high dimensional data. A typical approach for DML with high dimensional data is to perform the dimensionality reduction first before learning the distance metric. The main shortcoming of this approach is that it may result in a suboptimal sol...
Main Authors: | , , , |
---|---|
Format: | Report |
Language: | unknown |
Published: |
arXiv
2015
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.1509.04355 https://arxiv.org/abs/1509.04355 |
id |
ftdatacite:10.48550/arxiv.1509.04355 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.1509.04355 2023-05-15T16:01:12+02:00 Towards Making High Dimensional Distance Metric Learning Practical Qian, Qi Jin, Rong Zhang, Lijun Zhu, Shenghuo 2015 https://dx.doi.org/10.48550/arxiv.1509.04355 https://arxiv.org/abs/1509.04355 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Machine Learning cs.LG FOS Computer and information sciences Preprint Article article CreativeWork 2015 ftdatacite https://doi.org/10.48550/arxiv.1509.04355 2022-04-01T11:50:50Z In this work, we study distance metric learning (DML) for high dimensional data. A typical approach for DML with high dimensional data is to perform the dimensionality reduction first before learning the distance metric. The main shortcoming of this approach is that it may result in a suboptimal solution due to the subspace removed by the dimensionality reduction method. In this work, we present a dual random projection frame for DML with high dimensional data that explicitly addresses the limitation of dimensionality reduction for DML. The key idea is to first project all the data points into a low dimensional space by random projection, and compute the dual variables using the projected vectors. It then reconstructs the distance metric in the original space using the estimated dual variables. The proposed method, on one hand, enjoys the light computation of random projection, and on the other hand, alleviates the limitation of most dimensionality reduction methods. We verify both empirically and theoretically the effectiveness of the proposed algorithm for high dimensional DML. Report DML DataCite Metadata Store (German National Library of Science and Technology) |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
Machine Learning cs.LG FOS Computer and information sciences |
spellingShingle |
Machine Learning cs.LG FOS Computer and information sciences Qian, Qi Jin, Rong Zhang, Lijun Zhu, Shenghuo Towards Making High Dimensional Distance Metric Learning Practical |
topic_facet |
Machine Learning cs.LG FOS Computer and information sciences |
description |
In this work, we study distance metric learning (DML) for high dimensional data. A typical approach for DML with high dimensional data is to perform the dimensionality reduction first before learning the distance metric. The main shortcoming of this approach is that it may result in a suboptimal solution due to the subspace removed by the dimensionality reduction method. In this work, we present a dual random projection frame for DML with high dimensional data that explicitly addresses the limitation of dimensionality reduction for DML. The key idea is to first project all the data points into a low dimensional space by random projection, and compute the dual variables using the projected vectors. It then reconstructs the distance metric in the original space using the estimated dual variables. The proposed method, on one hand, enjoys the light computation of random projection, and on the other hand, alleviates the limitation of most dimensionality reduction methods. We verify both empirically and theoretically the effectiveness of the proposed algorithm for high dimensional DML. |
format |
Report |
author |
Qian, Qi Jin, Rong Zhang, Lijun Zhu, Shenghuo |
author_facet |
Qian, Qi Jin, Rong Zhang, Lijun Zhu, Shenghuo |
author_sort |
Qian, Qi |
title |
Towards Making High Dimensional Distance Metric Learning Practical |
title_short |
Towards Making High Dimensional Distance Metric Learning Practical |
title_full |
Towards Making High Dimensional Distance Metric Learning Practical |
title_fullStr |
Towards Making High Dimensional Distance Metric Learning Practical |
title_full_unstemmed |
Towards Making High Dimensional Distance Metric Learning Practical |
title_sort |
towards making high dimensional distance metric learning practical |
publisher |
arXiv |
publishDate |
2015 |
url |
https://dx.doi.org/10.48550/arxiv.1509.04355 https://arxiv.org/abs/1509.04355 |
genre |
DML |
genre_facet |
DML |
op_rights |
arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
op_doi |
https://doi.org/10.48550/arxiv.1509.04355 |
_version_ |
1766397165130743808 |