Totally-corrective multi-class boosting

Zhihui Hao; Chunhua Shen; Nick Barnes; Bo Wang

doi:10.1007/978-3-642-19282-1_22

Totally-corrective multi-class boosting

Zhihui Hao^*, Chunhua Shen, Nick Barnes, Bo Wang

^*Corresponding author for this work

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

3 Citations (Scopus)

Abstract

We proffer totally-corrective multi-class boosting algorithms in this work. First, we discuss the methods that extend two-class boosting to multi-class case by studying two existing boosting algorithms: AdaBoost.MO and SAMME, and formulate convex optimization problems that minimize their regularized cost functions. Then we propose a column-generation based totally-corrective framework for multi-class boosting learning by looking at the Lagrange dual problems. Experimental results on UCI datasets show that the new algorithms have comparable generalization capability but converge much faster than their counterparts. Experiments on MNIST handwriting digit classification also demonstrate the effectiveness of the proposed algorithms.

Original language	English
Title of host publication	Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers
Pages	269-280
Number of pages	12
Edition	PART 4
DOIs	https://doi.org/10.1007/978-3-642-19282-1_22
Publication status	Published - 2011
Event	10th Asian Conference on Computer Vision, ACCV 2010 - Queenstown, New Zealand Duration: 8 Nov 2010 → 12 Nov 2010 https://link.springer.com/book/10.1007/978-3-642-19282-1

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 4
Volume	6495 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	10th Asian Conference on Computer Vision, ACCV 2010
Country/Territory	New Zealand
City	Queenstown
Period	8/11/10 → 12/11/10
Internet address	https://link.springer.com/book/10.1007/978-3-642-19282-1

Access to Document

10.1007/978-3-642-19282-1_22

Cite this

Hao, Z., Shen, C., Barnes, N., & Wang, B. (2011). Totally-corrective multi-class boosting. In Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers (PART 4 ed., pp. 269-280). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6495 LNCS, No. PART 4). https://doi.org/10.1007/978-3-642-19282-1_22

@inproceedings{d4524022012a4b848130ddd74ab4ae2a,

title = "Totally-corrective multi-class boosting",

abstract = "We proffer totally-corrective multi-class boosting algorithms in this work. First, we discuss the methods that extend two-class boosting to multi-class case by studying two existing boosting algorithms: AdaBoost.MO and SAMME, and formulate convex optimization problems that minimize their regularized cost functions. Then we propose a column-generation based totally-corrective framework for multi-class boosting learning by looking at the Lagrange dual problems. Experimental results on UCI datasets show that the new algorithms have comparable generalization capability but converge much faster than their counterparts. Experiments on MNIST handwriting digit classification also demonstrate the effectiveness of the proposed algorithms.",

author = "Zhihui Hao and Chunhua Shen and Nick Barnes and Bo Wang",

year = "2011",

doi = "10.1007/978-3-642-19282-1_22",

language = "English",

isbn = "9783642192814",

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

number = "PART 4",

pages = "269--280",

booktitle = "Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers",

edition = "PART 4",

note = "10th Asian Conference on Computer Vision, ACCV 2010 ; Conference date: 08-11-2010 Through 12-11-2010",

url = "https://link.springer.com/book/10.1007/978-3-642-19282-1",

}

Hao, Z, Shen, C, Barnes, N & Wang, B 2011, Totally-corrective multi-class boosting. in Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 4 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 4, vol. 6495 LNCS, pp. 269-280, 10th Asian Conference on Computer Vision, ACCV 2010, Queenstown, New Zealand, 8/11/10. https://doi.org/10.1007/978-3-642-19282-1_22

Totally-corrective multi-class boosting. / Hao, Zhihui; Shen, Chunhua; Barnes, Nick et al.
Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 4. ed. 2011. p. 269-280 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6495 LNCS, No. PART 4).

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

TY - GEN

T1 - Totally-corrective multi-class boosting

AU - Hao, Zhihui

AU - Shen, Chunhua

AU - Barnes, Nick

AU - Wang, Bo

PY - 2011

Y1 - 2011

N2 - We proffer totally-corrective multi-class boosting algorithms in this work. First, we discuss the methods that extend two-class boosting to multi-class case by studying two existing boosting algorithms: AdaBoost.MO and SAMME, and formulate convex optimization problems that minimize their regularized cost functions. Then we propose a column-generation based totally-corrective framework for multi-class boosting learning by looking at the Lagrange dual problems. Experimental results on UCI datasets show that the new algorithms have comparable generalization capability but converge much faster than their counterparts. Experiments on MNIST handwriting digit classification also demonstrate the effectiveness of the proposed algorithms.

AB - We proffer totally-corrective multi-class boosting algorithms in this work. First, we discuss the methods that extend two-class boosting to multi-class case by studying two existing boosting algorithms: AdaBoost.MO and SAMME, and formulate convex optimization problems that minimize their regularized cost functions. Then we propose a column-generation based totally-corrective framework for multi-class boosting learning by looking at the Lagrange dual problems. Experimental results on UCI datasets show that the new algorithms have comparable generalization capability but converge much faster than their counterparts. Experiments on MNIST handwriting digit classification also demonstrate the effectiveness of the proposed algorithms.

UR - http://www.scopus.com/inward/record.url?scp=79952532175&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-19282-1_22

DO - 10.1007/978-3-642-19282-1_22

M3 - Conference contribution

SN - 9783642192814

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

SP - 269

EP - 280

BT - Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers

T2 - 10th Asian Conference on Computer Vision, ACCV 2010

Y2 - 8 November 2010 through 12 November 2010

ER -

Hao Z, Shen C, Barnes N, Wang B. Totally-corrective multi-class boosting. In Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 4 ed. 2011. p. 269-280. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 4). doi: 10.1007/978-3-642-19282-1_22

Totally-corrective multi-class boosting

Abstract

Publication series

Conference

Access to Document

Other files and links

Cite this