Exploiting side information in locality preserving projection

Even if the class label information is unknown, side information represents some equivalence constraints between pairs of patterns, indicating whether pairs originate from the same class. Exploiting side information, we develop algorithms to preserve both the intra-class and inter-class local struct...

Full description

Bibliographic Details
Main Authors: An, Senjian, Liu, Wan-Quan, Venkatesh, Svetha
Other Authors: NA
Format: Conference Paper
Published: IEEE 2008
Online Access:http://hdl.handle.net/20.500.11937/30673
_version_ 1848753154815950848
author An, Senjian
Liu, Wan-Quan
Venkatesh, Svetha
author2 NA
author_facet NA
An, Senjian
Liu, Wan-Quan
Venkatesh, Svetha
author_sort An, Senjian
building Curtin Institutional Repository
collection Online Access
description Even if the class label information is unknown, side information represents some equivalence constraints between pairs of patterns, indicating whether pairs originate from the same class. Exploiting side information, we develop algorithms to preserve both the intra-class and inter-class local structures. This new type of locality preserving projection (LPP), called LPP with side information (LPPSI), preserves the datapsilas local structure in the sense that the close, similar training patterns will be kept close, whilst the close but dissimilar ones are separated. Our algorithms balance these conflicting requirements, and we further improve this technique using kernel methods. Experiments conducted on popular face databases demonstrate that the proposed algorithm significantly outperforms LPP. Further, we show that the performance of our algorithm with partial side information (that is, using only small amount of pair-wise similarity/dissimilarity information during training) is comparable with that when using full side information. We conclude that exploiting side information by preserving both similar and dissimilar local structures of the data significantly improves performance.
first_indexed 2025-11-14T08:20:01Z
format Conference Paper
id curtin-20.500.11937-30673
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T08:20:01Z
publishDate 2008
publisher IEEE
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-306732018-03-29T09:08:00Z Exploiting side information in locality preserving projection An, Senjian Liu, Wan-Quan Venkatesh, Svetha NA Even if the class label information is unknown, side information represents some equivalence constraints between pairs of patterns, indicating whether pairs originate from the same class. Exploiting side information, we develop algorithms to preserve both the intra-class and inter-class local structures. This new type of locality preserving projection (LPP), called LPP with side information (LPPSI), preserves the datapsilas local structure in the sense that the close, similar training patterns will be kept close, whilst the close but dissimilar ones are separated. Our algorithms balance these conflicting requirements, and we further improve this technique using kernel methods. Experiments conducted on popular face databases demonstrate that the proposed algorithm significantly outperforms LPP. Further, we show that the performance of our algorithm with partial side information (that is, using only small amount of pair-wise similarity/dissimilarity information during training) is comparable with that when using full side information. We conclude that exploiting side information by preserving both similar and dissimilar local structures of the data significantly improves performance. 2008 Conference Paper http://hdl.handle.net/20.500.11937/30673 10.1109/CVPR.2008.4587596 IEEE restricted
spellingShingle An, Senjian
Liu, Wan-Quan
Venkatesh, Svetha
Exploiting side information in locality preserving projection
title Exploiting side information in locality preserving projection
title_full Exploiting side information in locality preserving projection
title_fullStr Exploiting side information in locality preserving projection
title_full_unstemmed Exploiting side information in locality preserving projection
title_short Exploiting side information in locality preserving projection
title_sort exploiting side information in locality preserving projection
url http://hdl.handle.net/20.500.11937/30673