Twin Kernel Embedding with Back Constraints

Yi Guo; Paul Hing Kwan; Junbin Gao

doi:10.1109/ICDMW.2007.15

Twin Kernel Embedding with Back Constraints

Yi Guo, Paul Hing Kwan, Junbin Gao

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Citations (Scopus)

Abstract

Twin kernel embedding (TKE) is a novel approach for visualization of non-vectorial objects. It preserves the similarity structure in high-dimensional or structured input data and reproduces it in a low dimensional latent space by matching the similarity relations represented by two kernel gram matrices, one kernel for the input data and the other for embedded data. However, there is no explicit mapping from the input data to their corresponding low dimensional embeddings. We obtain this mapping by including the back constraints on the data in TKE in this paper. This procedure still emphasizes the locality preserving. Further, the smooth mapping also solves the problem of so-called out-of-sample problem which is absent in the original TKE. Experimental evaluation on different real world data sets verifies the usefulness of this method.

Original language	English
Title of host publication	Proceedings of the Seventh IEEE International Conference on Data Mining Workshops
Editors	IEEE: Institute of Electrical, Electronics Engineers Systems Computer Society
Place of Publication	Los Alamitos, United States of America
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	319-324
ISBN (Print)	0769530338
DOIs	https://doi.org/10.1109/ICDMW.2007.15
Publication status	Published - 2007
Event	ICDMW 2007: Seventh IEEE International Conference on Data Mining Workshops - Omaha, United States of America Duration: 28 Oct 2007 → 31 Oct 2007

Conference

Conference	ICDMW 2007: Seventh IEEE International Conference on Data Mining Workshops
City	Omaha, United States of America
Period	28/10/07 → 31/10/07

Keywords

Pattern Recognition and Data Mining

Access to Document

10.1109/ICDMW.2007.15

https://hdl.handle.net/1959.11/4767

Handle.net

Persistent link

Cite this

Guo, Yi ; Kwan, Paul Hing ; Gao, Junbin. / Twin Kernel Embedding with Back Constraints. Proceedings of the Seventh IEEE International Conference on Data Mining Workshops. editor / IEEE: Institute of Electrical ; Electronics Engineers Systems Computer Society. Los Alamitos, United States of America : Institute of Electrical and Electronics Engineers (IEEE), 2007. pp. 319-324

@inproceedings{9d585bde238646d7b2e1a4a5dc474b4d,

title = "Twin Kernel Embedding with Back Constraints",

abstract = "Twin kernel embedding (TKE) is a novel approach for visualization of non-vectorial objects. It preserves the similarity structure in high-dimensional or structured input data and reproduces it in a low dimensional latent space by matching the similarity relations represented by two kernel gram matrices, one kernel for the input data and the other for embedded data. However, there is no explicit mapping from the input data to their corresponding low dimensional embeddings. We obtain this mapping by including the back constraints on the data in TKE in this paper. This procedure still emphasizes the locality preserving. Further, the smooth mapping also solves the problem of so-called out-of-sample problem which is absent in the original TKE. Experimental evaluation on different real world data sets verifies the usefulness of this method.",

keywords = "Pattern Recognition and Data Mining",

author = "Yi Guo and Kwan, \{Paul Hing\} and Junbin Gao",

year = "2007",

doi = "10.1109/ICDMW.2007.15",

language = "English",

isbn = "0769530338",

pages = "319--324",

editor = "\{Institute of Electrical\}, IEEE: and \{Engineers Systems Computer Society\}, Electronics",

booktitle = "Proceedings of the Seventh IEEE International Conference on Data Mining Workshops",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

note = "ICDMW 2007: Seventh IEEE International Conference on Data Mining Workshops ; Conference date: 28-10-2007 Through 31-10-2007",

}

Guo, Y, Kwan, PH & Gao, J 2007, Twin Kernel Embedding with Back Constraints. in IEEE Institute of Electrical & E Engineers Systems Computer Society (eds), Proceedings of the Seventh IEEE International Conference on Data Mining Workshops. Institute of Electrical and Electronics Engineers (IEEE), Los Alamitos, United States of America, pp. 319-324, ICDMW 2007: Seventh IEEE International Conference on Data Mining Workshops, Omaha, United States of America, 28/10/07. https://doi.org/10.1109/ICDMW.2007.15

Twin Kernel Embedding with Back Constraints. / Guo, Yi; Kwan, Paul Hing; Gao, Junbin.
Proceedings of the Seventh IEEE International Conference on Data Mining Workshops. ed. / IEEE: Institute of Electrical; Electronics Engineers Systems Computer Society. Los Alamitos, United States of America: Institute of Electrical and Electronics Engineers (IEEE), 2007. p. 319-324.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Twin Kernel Embedding with Back Constraints

AU - Guo, Yi

AU - Kwan, Paul Hing

AU - Gao, Junbin

PY - 2007

Y1 - 2007

N2 - Twin kernel embedding (TKE) is a novel approach for visualization of non-vectorial objects. It preserves the similarity structure in high-dimensional or structured input data and reproduces it in a low dimensional latent space by matching the similarity relations represented by two kernel gram matrices, one kernel for the input data and the other for embedded data. However, there is no explicit mapping from the input data to their corresponding low dimensional embeddings. We obtain this mapping by including the back constraints on the data in TKE in this paper. This procedure still emphasizes the locality preserving. Further, the smooth mapping also solves the problem of so-called out-of-sample problem which is absent in the original TKE. Experimental evaluation on different real world data sets verifies the usefulness of this method.

AB - Twin kernel embedding (TKE) is a novel approach for visualization of non-vectorial objects. It preserves the similarity structure in high-dimensional or structured input data and reproduces it in a low dimensional latent space by matching the similarity relations represented by two kernel gram matrices, one kernel for the input data and the other for embedded data. However, there is no explicit mapping from the input data to their corresponding low dimensional embeddings. We obtain this mapping by including the back constraints on the data in TKE in this paper. This procedure still emphasizes the locality preserving. Further, the smooth mapping also solves the problem of so-called out-of-sample problem which is absent in the original TKE. Experimental evaluation on different real world data sets verifies the usefulness of this method.

KW - Pattern Recognition and Data Mining

UR - https://www.scopus.com/pages/publications/46149095008

U2 - 10.1109/ICDMW.2007.15

DO - 10.1109/ICDMW.2007.15

M3 - Conference contribution

SN - 0769530338

SP - 319

EP - 324

BT - Proceedings of the Seventh IEEE International Conference on Data Mining Workshops

A2 - Institute of Electrical, IEEE:

A2 - Engineers Systems Computer Society, Electronics

PB - Institute of Electrical and Electronics Engineers (IEEE)

CY - Los Alamitos, United States of America

T2 - ICDMW 2007: Seventh IEEE International Conference on Data Mining Workshops

Y2 - 28 October 2007 through 31 October 2007

ER -

Guo Y, Kwan PH, Gao J. Twin Kernel Embedding with Back Constraints. In Institute of Electrical IEEE, Engineers Systems Computer Society E, editors, Proceedings of the Seventh IEEE International Conference on Data Mining Workshops. Los Alamitos, United States of America: Institute of Electrical and Electronics Engineers (IEEE). 2007. p. 319-324 doi: 10.1109/ICDMW.2007.15

Twin Kernel Embedding with Back Constraints

Abstract

Conference

Keywords

Access to Document

Handle.net

Fingerprint

Cite this