Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD)
Distance metric learning (DML) is an important task that has found applications in many domains. The high computational cost of DML arises from the large number of variables to be determined and the constraint that a distance metric has to be a positive semi-definite (PSD) matrix. Although stochasti...
Main Authors: | , , , , |
---|---|
Format: | Report |
Language: | unknown |
Published: |
arXiv
2013
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.1304.1192 https://arxiv.org/abs/1304.1192 |
id |
ftdatacite:10.48550/arxiv.1304.1192 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.1304.1192 2023-05-15T16:01:10+02:00 Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) Qian, Qi Jin, Rong Yi, Jinfeng Zhang, Lijun Zhu, Shenghuo 2013 https://dx.doi.org/10.48550/arxiv.1304.1192 https://arxiv.org/abs/1304.1192 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 2013 ftdatacite https://doi.org/10.48550/arxiv.1304.1192 2022-04-01T13:23:02Z Distance metric learning (DML) is an important task that has found applications in many domains. The high computational cost of DML arises from the large number of variables to be determined and the constraint that a distance metric has to be a positive semi-definite (PSD) matrix. Although stochastic gradient descent (SGD) has been successfully applied to improve the efficiency of DML, it can still be computationally expensive because in order to ensure that the solution is a PSD matrix, it has to, at every iteration, project the updated distance metric onto the PSD cone, an expensive operation. We address this challenge by developing two strategies within SGD, i.e. mini-batch and adaptive sampling, to effectively reduce the number of updates (i.e., projections onto the PSD cone) in SGD. We also develop hybrid approaches that combine the strength of adaptive sampling with that of mini-batch online learning techniques to further improve the computational efficiency of SGD for DML. We prove the theoretical guarantees for both adaptive sampling and mini-batch based approaches for DML. We also conduct an extensive empirical study to verify the effectiveness of the proposed algorithms for 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 Yi, Jinfeng Zhang, Lijun Zhu, Shenghuo Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
topic_facet |
Machine Learning cs.LG FOS Computer and information sciences |
description |
Distance metric learning (DML) is an important task that has found applications in many domains. The high computational cost of DML arises from the large number of variables to be determined and the constraint that a distance metric has to be a positive semi-definite (PSD) matrix. Although stochastic gradient descent (SGD) has been successfully applied to improve the efficiency of DML, it can still be computationally expensive because in order to ensure that the solution is a PSD matrix, it has to, at every iteration, project the updated distance metric onto the PSD cone, an expensive operation. We address this challenge by developing two strategies within SGD, i.e. mini-batch and adaptive sampling, to effectively reduce the number of updates (i.e., projections onto the PSD cone) in SGD. We also develop hybrid approaches that combine the strength of adaptive sampling with that of mini-batch online learning techniques to further improve the computational efficiency of SGD for DML. We prove the theoretical guarantees for both adaptive sampling and mini-batch based approaches for DML. We also conduct an extensive empirical study to verify the effectiveness of the proposed algorithms for DML. |
format |
Report |
author |
Qian, Qi Jin, Rong Yi, Jinfeng Zhang, Lijun Zhu, Shenghuo |
author_facet |
Qian, Qi Jin, Rong Yi, Jinfeng Zhang, Lijun Zhu, Shenghuo |
author_sort |
Qian, Qi |
title |
Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
title_short |
Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
title_full |
Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
title_fullStr |
Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
title_full_unstemmed |
Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD) |
title_sort |
efficient distance metric learning by adaptive sampling and mini-batch stochastic gradient descent (sgd) |
publisher |
arXiv |
publishDate |
2013 |
url |
https://dx.doi.org/10.48550/arxiv.1304.1192 https://arxiv.org/abs/1304.1192 |
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.1304.1192 |
_version_ |
1766397141799927808 |