Can AdaBoost.M1 learn incrementally? A comparison to learn++ under different combination rules

Hussein Syed Mohammed; James Leander; Matthew Marbach; Robi Polikar

doi:10.1007/11840817_27

Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules

Hussein Syed Mohammed, James Leander, Matthew Marbach, Robi Polikar

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

6 Scopus citations

Abstract

We had previously introduced Learn⁺⁺, inspired in part by the ensemble based AdaBoost algorithm, for incrementally learning from new data, including new concept classes, without forgetting what had been previously learned. In this effort, we compare the incremental learning performance of Learn⁺⁺ and AdaBoost under several combination schemes, including their native, weighted majority voting. We show on several databases that changing AdaBoost's distribution update rule from hypothesis based update to ensemble based update allows significantly more efficient incremental learning ability, regardless of the combination rule used to combine the classifiers.

Original language	English (US)
Title of host publication	Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings
Publisher	Springer Verlag
Pages	254-263
Number of pages	10
ISBN (Print)	3540386254, 9783540386254
DOIs	https://doi.org/10.1007/11840817_27
State	Published - 2006
Externally published	Yes
Event	16th International Conference on Artificial Neural Networks, ICANN 2006 - Athens, Greece Duration: Sep 10 2006 → Sep 14 2006

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	4131 LNCS - I
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	16th International Conference on Artificial Neural Networks, ICANN 2006
Country/Territory	Greece
City	Athens
Period	9/10/06 → 9/14/06

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/11840817_27

Cite this

Mohammed, H. S., Leander, J., Marbach, M., & Polikar, R. (2006). Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules. In Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings (pp. 254-263). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4131 LNCS - I). Springer Verlag. https://doi.org/10.1007/11840817_27

Mohammed, Hussein Syed ; Leander, James ; Marbach, Matthew et al. / Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules. Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings. Springer Verlag, 2006. pp. 254-263 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{372d2264ddd0461a8e0a14bbc76a1171,

title = "Can AdaBoost.M1 learn incrementally? A comparison to learn++ under different combination rules",

abstract = "We had previously introduced Learn++, inspired in part by the ensemble based AdaBoost algorithm, for incrementally learning from new data, including new concept classes, without forgetting what had been previously learned. In this effort, we compare the incremental learning performance of Learn++ and AdaBoost under several combination schemes, including their native, weighted majority voting. We show on several databases that changing AdaBoost's distribution update rule from hypothesis based update to ensemble based update allows significantly more efficient incremental learning ability, regardless of the combination rule used to combine the classifiers.",

author = "Mohammed, {Hussein Syed} and James Leander and Matthew Marbach and Robi Polikar",

year = "2006",

doi = "10.1007/11840817_27",

language = "English (US)",

isbn = "3540386254",

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

publisher = "Springer Verlag",

pages = "254--263",

booktitle = "Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings",

address = "Germany",

note = "16th International Conference on Artificial Neural Networks, ICANN 2006 ; Conference date: 10-09-2006 Through 14-09-2006",

}

Mohammed, HS, Leander, J, Marbach, M & Polikar, R 2006, Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules. in Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4131 LNCS - I, Springer Verlag, pp. 254-263, 16th International Conference on Artificial Neural Networks, ICANN 2006, Athens, Greece, 9/10/06. https://doi.org/10.1007/11840817_27

Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules. / Mohammed, Hussein Syed; Leander, James; Marbach, Matthew et al.
Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings. Springer Verlag, 2006. p. 254-263 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4131 LNCS - I).

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

TY - GEN

T1 - Can AdaBoost.M1 learn incrementally? A comparison to learn++ under different combination rules

AU - Mohammed, Hussein Syed

AU - Leander, James

AU - Marbach, Matthew

AU - Polikar, Robi

PY - 2006

Y1 - 2006

N2 - We had previously introduced Learn++, inspired in part by the ensemble based AdaBoost algorithm, for incrementally learning from new data, including new concept classes, without forgetting what had been previously learned. In this effort, we compare the incremental learning performance of Learn++ and AdaBoost under several combination schemes, including their native, weighted majority voting. We show on several databases that changing AdaBoost's distribution update rule from hypothesis based update to ensemble based update allows significantly more efficient incremental learning ability, regardless of the combination rule used to combine the classifiers.

AB - We had previously introduced Learn++, inspired in part by the ensemble based AdaBoost algorithm, for incrementally learning from new data, including new concept classes, without forgetting what had been previously learned. In this effort, we compare the incremental learning performance of Learn++ and AdaBoost under several combination schemes, including their native, weighted majority voting. We show on several databases that changing AdaBoost's distribution update rule from hypothesis based update to ensemble based update allows significantly more efficient incremental learning ability, regardless of the combination rule used to combine the classifiers.

UR - http://www.scopus.com/inward/record.url?scp=33749842512&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33749842512&partnerID=8YFLogxK

U2 - 10.1007/11840817_27

DO - 10.1007/11840817_27

M3 - Conference contribution

AN - SCOPUS:33749842512

SN - 3540386254

SN - 9783540386254

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

SP - 254

EP - 263

BT - Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings

PB - Springer Verlag

T2 - 16th International Conference on Artificial Neural Networks, ICANN 2006

Y2 - 10 September 2006 through 14 September 2006

ER -

Mohammed HS, Leander J, Marbach M, Polikar R. Can AdaBoost.M1 learn incrementally? A comparison to learn⁺⁺ under different combination rules. In Artificial Neural Networks, ICANN 2006 - 16th International Conference, Proceedings. Springer Verlag. 2006. p. 254-263. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/11840817_27