Kernel-based Distance Metric Learning in the Output Space

In this paper we present two related, kernel-based Distance Metric Learning (DML) methods. Their respective models non-linearly map data from their original space to an output space, and subsequent distance measurements are performed in the output space via a Mahalanobis metric. The dimensionality o...

Full description

Bibliographic Details
Main Authors: Li, Cong, Georgiopoulos, Michael, Anagnostopoulos, Georgios C.
Format: Text
Language:unknown
Published: arXiv 2013
Subjects:
Machine Learning cs.LG
FOS Computer and information sciences
DML
Online Access:https://dx.doi.org/10.48550/arxiv.1312.2578
https://arxiv.org/abs/1312.2578
id ftdatacite:10.48550/arxiv.1312.2578
record_format openpolar
spelling ftdatacite:10.48550/arxiv.1312.2578 2023-05-15T16:01:32+02:00 Kernel-based Distance Metric Learning in the Output Space Li, Cong Georgiopoulos, Michael Anagnostopoulos, Georgios C. 2013 https://dx.doi.org/10.48550/arxiv.1312.2578 https://arxiv.org/abs/1312.2578 unknown arXiv https://dx.doi.org/10.1109/ijcnn.2013.6706862 arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Machine Learning cs.LG FOS Computer and information sciences article-journal Article ScholarlyArticle Text 2013 ftdatacite https://doi.org/10.48550/arxiv.1312.2578 https://doi.org/10.1109/ijcnn.2013.6706862 2022-04-01T13:06:23Z In this paper we present two related, kernel-based Distance Metric Learning (DML) methods. Their respective models non-linearly map data from their original space to an output space, and subsequent distance measurements are performed in the output space via a Mahalanobis metric. The dimensionality of the output space can be directly controlled to facilitate the learning of a low-rank metric. Both methods allow for simultaneous inference of the associated metric and the mapping to the output space, which can be used to visualize the data, when the output space is 2- or 3-dimensional. Experimental results for a collection of classification tasks illustrate the advantages of the proposed methods over other traditional and kernel-based DML approaches. : 11 pages, 7 figures, appeared in the Proceedings of 2013 International Joint Conference on Neural Networks (IJCNN) Text 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
Li, Cong
Georgiopoulos, Michael
Anagnostopoulos, Georgios C.
Kernel-based Distance Metric Learning in the Output Space
topic_facet Machine Learning cs.LG
FOS Computer and information sciences
description In this paper we present two related, kernel-based Distance Metric Learning (DML) methods. Their respective models non-linearly map data from their original space to an output space, and subsequent distance measurements are performed in the output space via a Mahalanobis metric. The dimensionality of the output space can be directly controlled to facilitate the learning of a low-rank metric. Both methods allow for simultaneous inference of the associated metric and the mapping to the output space, which can be used to visualize the data, when the output space is 2- or 3-dimensional. Experimental results for a collection of classification tasks illustrate the advantages of the proposed methods over other traditional and kernel-based DML approaches. : 11 pages, 7 figures, appeared in the Proceedings of 2013 International Joint Conference on Neural Networks (IJCNN)
format Text
author Li, Cong
Georgiopoulos, Michael
Anagnostopoulos, Georgios C.
author_facet Li, Cong
Georgiopoulos, Michael
Anagnostopoulos, Georgios C.
author_sort Li, Cong
title Kernel-based Distance Metric Learning in the Output Space
title_short Kernel-based Distance Metric Learning in the Output Space
title_full Kernel-based Distance Metric Learning in the Output Space
title_fullStr Kernel-based Distance Metric Learning in the Output Space
title_full_unstemmed Kernel-based Distance Metric Learning in the Output Space
title_sort kernel-based distance metric learning in the output space
publisher arXiv
publishDate 2013
url https://dx.doi.org/10.48550/arxiv.1312.2578
https://arxiv.org/abs/1312.2578
genre DML
genre_facet DML
op_relation https://dx.doi.org/10.1109/ijcnn.2013.6706862
op_rights arXiv.org perpetual, non-exclusive license
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
op_doi https://doi.org/10.48550/arxiv.1312.2578
https://doi.org/10.1109/ijcnn.2013.6706862
_version_ 1766397352565800960

Similar Items