Constructing sparse KFDA using pre-image reconstruction

Qing Zhang; Jianwu Li

doi:10.1007/978-3-642-17534-3_81

Constructing sparse KFDA using pre-image reconstruction

Qing Zhang, Jianwu Li^*

^*Corresponding author for this work

School of Computer Science and Technology

Institute of Scientific and Technical Information of China

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

3 Citations (Scopus)

Abstract

Kernel Fisher Discriminant Analysis (KFDA) improves greatly the classification accuracy of FDA via using kernel trick. However, the final solution of KFDA is expressed as an expansion of all training examples, which seriously undermines the classification efficiency, especially in real-time applications. This paper proposes a novel framework to construct sparse KFDA using pre-image reconstruction. The proposed method (PR-KFDA) appends greedily the pre-image of the residual between the current approximate model and the original KFDA model in feature space with the local optimal Fisher coefficients to acquire sparse representation of KFDA solution. Experimental results show that PR-KFDA can reduce the solution of KFDA effectively while maintaining comparable test accuracy.

Original language	English
Title of host publication	Neural Information Processing
Subtitle of host publication	Models and Applications - 17th International Conference, ICONIP 2010, Proceedings
Pages	658-667
Number of pages	10
Edition	PART 2
DOIs	https://doi.org/10.1007/978-3-642-17534-3_81
Publication status	Published - 2010
Event	17th International Conference on Neural Information Processing, ICONIP 2010 - Sydney, NSW, Australia Duration: 22 Nov 2010 → 25 Nov 2010

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 2
Volume	6444 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	17th International Conference on Neural Information Processing, ICONIP 2010
Country/Territory	Australia
City	Sydney, NSW
Period	22/11/10 → 25/11/10

Keywords

KFDA
kernel method
pre-image reconstruction
sparse approximation framework

Access to Document

10.1007/978-3-642-17534-3_81

Cite this

Zhang, Q., & Li, J. (2010). Constructing sparse KFDA using pre-image reconstruction. In Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings (PART 2 ed., pp. 658-667). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6444 LNCS, No. PART 2). https://doi.org/10.1007/978-3-642-17534-3_81

Zhang, Qing ; Li, Jianwu. / Constructing sparse KFDA using pre-image reconstruction. Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2. ed. 2010. pp. 658-667 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).

@inproceedings{2ade9432ec8b44e9a544222e0ad07d8f,

title = "Constructing sparse KFDA using pre-image reconstruction",

abstract = "Kernel Fisher Discriminant Analysis (KFDA) improves greatly the classification accuracy of FDA via using kernel trick. However, the final solution of KFDA is expressed as an expansion of all training examples, which seriously undermines the classification efficiency, especially in real-time applications. This paper proposes a novel framework to construct sparse KFDA using pre-image reconstruction. The proposed method (PR-KFDA) appends greedily the pre-image of the residual between the current approximate model and the original KFDA model in feature space with the local optimal Fisher coefficients to acquire sparse representation of KFDA solution. Experimental results show that PR-KFDA can reduce the solution of KFDA effectively while maintaining comparable test accuracy.",

keywords = "KFDA, kernel method, pre-image reconstruction, sparse approximation framework",

author = "Qing Zhang and Jianwu Li",

year = "2010",

doi = "10.1007/978-3-642-17534-3_81",

language = "English",

isbn = "3642175333",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

number = "PART 2",

pages = "658--667",

booktitle = "Neural Information Processing",

edition = "PART 2",

note = "17th International Conference on Neural Information Processing, ICONIP 2010 ; Conference date: 22-11-2010 Through 25-11-2010",

}

Zhang, Q & Li, J 2010, Constructing sparse KFDA using pre-image reconstruction. in Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 6444 LNCS, pp. 658-667, 17th International Conference on Neural Information Processing, ICONIP 2010, Sydney, NSW, Australia, 22/11/10. https://doi.org/10.1007/978-3-642-17534-3_81

Constructing sparse KFDA using pre-image reconstruction. / Zhang, Qing; Li, Jianwu.
Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2. ed. 2010. p. 658-667 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6444 LNCS, No. PART 2).

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

TY - GEN

T1 - Constructing sparse KFDA using pre-image reconstruction

AU - Zhang, Qing

AU - Li, Jianwu

PY - 2010

Y1 - 2010

N2 - Kernel Fisher Discriminant Analysis (KFDA) improves greatly the classification accuracy of FDA via using kernel trick. However, the final solution of KFDA is expressed as an expansion of all training examples, which seriously undermines the classification efficiency, especially in real-time applications. This paper proposes a novel framework to construct sparse KFDA using pre-image reconstruction. The proposed method (PR-KFDA) appends greedily the pre-image of the residual between the current approximate model and the original KFDA model in feature space with the local optimal Fisher coefficients to acquire sparse representation of KFDA solution. Experimental results show that PR-KFDA can reduce the solution of KFDA effectively while maintaining comparable test accuracy.

AB - Kernel Fisher Discriminant Analysis (KFDA) improves greatly the classification accuracy of FDA via using kernel trick. However, the final solution of KFDA is expressed as an expansion of all training examples, which seriously undermines the classification efficiency, especially in real-time applications. This paper proposes a novel framework to construct sparse KFDA using pre-image reconstruction. The proposed method (PR-KFDA) appends greedily the pre-image of the residual between the current approximate model and the original KFDA model in feature space with the local optimal Fisher coefficients to acquire sparse representation of KFDA solution. Experimental results show that PR-KFDA can reduce the solution of KFDA effectively while maintaining comparable test accuracy.

KW - KFDA

KW - kernel method

KW - pre-image reconstruction

KW - sparse approximation framework

UR - http://www.scopus.com/inward/record.url?scp=78650214366&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-17534-3_81

DO - 10.1007/978-3-642-17534-3_81

M3 - Conference contribution

AN - SCOPUS:78650214366

SN - 3642175333

SN - 9783642175336

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 658

EP - 667

BT - Neural Information Processing

T2 - 17th International Conference on Neural Information Processing, ICONIP 2010

Y2 - 22 November 2010 through 25 November 2010

ER -

Zhang Q, Li J. Constructing sparse KFDA using pre-image reconstruction. In Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2 ed. 2010. p. 658-667. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). doi: 10.1007/978-3-642-17534-3_81

Constructing sparse KFDA using pre-image reconstruction

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this