Context-dependent fusion of multiple algorithms with minimum classification error learning

Lijun Zhang, Hichem Frigui, Paul Gader

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We present a novel method for fusing the decisions of multiple classification algorithms which use different features, classification methods, and data sources. The proposed method, called Context Dependent Fusion of Multiple Algorithms(CDF-MA) is motivated by the fact that the relative performance of different algorithms can vary significantly as the characteristics of the input data vary. The training part of CDF-MA has two main components: context extraction and algorithm fusion. In context extraction, the features used by the distinct algorithms are combined and clustered into groups of similar contexts. The algorithm fusion component embeds a variation of the MCE/GPD method to assigns an aggregation weight to each algorithm based on its loss function within each context. Results on real world data show that the proposed method can identify meaningful and coherent clusters, and outperform all individual classifiers and the global weighted average fusion method.

Original languageEnglish (US)
Title of host publication8th International Conference on Machine Learning and Applications, ICMLA 2009
Pages190-195
Number of pages6
DOIs
StatePublished - Dec 1 2009
Event8th International Conference on Machine Learning and Applications, ICMLA 2009 - Miami Beach, FL, United States
Duration: Dec 13 2009Dec 15 2009

Publication series

Name8th International Conference on Machine Learning and Applications, ICMLA 2009

Other

Other8th International Conference on Machine Learning and Applications, ICMLA 2009
CountryUnited States
CityMiami Beach, FL
Period12/13/0912/15/09

Fingerprint

Fusion reactions
Classifiers
Agglomeration

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Human-Computer Interaction
  • Software

Cite this

Zhang, L., Frigui, H., & Gader, P. (2009). Context-dependent fusion of multiple algorithms with minimum classification error learning. In 8th International Conference on Machine Learning and Applications, ICMLA 2009 (pp. 190-195). [5381831] (8th International Conference on Machine Learning and Applications, ICMLA 2009). https://doi.org/10.1109/ICMLA.2009.119
Zhang, Lijun ; Frigui, Hichem ; Gader, Paul. / Context-dependent fusion of multiple algorithms with minimum classification error learning. 8th International Conference on Machine Learning and Applications, ICMLA 2009. 2009. pp. 190-195 (8th International Conference on Machine Learning and Applications, ICMLA 2009).
@inproceedings{ec6ad8ed40104dcbb16931e171b5113a,
title = "Context-dependent fusion of multiple algorithms with minimum classification error learning",
abstract = "We present a novel method for fusing the decisions of multiple classification algorithms which use different features, classification methods, and data sources. The proposed method, called Context Dependent Fusion of Multiple Algorithms(CDF-MA) is motivated by the fact that the relative performance of different algorithms can vary significantly as the characteristics of the input data vary. The training part of CDF-MA has two main components: context extraction and algorithm fusion. In context extraction, the features used by the distinct algorithms are combined and clustered into groups of similar contexts. The algorithm fusion component embeds a variation of the MCE/GPD method to assigns an aggregation weight to each algorithm based on its loss function within each context. Results on real world data show that the proposed method can identify meaningful and coherent clusters, and outperform all individual classifiers and the global weighted average fusion method.",
author = "Lijun Zhang and Hichem Frigui and Paul Gader",
year = "2009",
month = "12",
day = "1",
doi = "10.1109/ICMLA.2009.119",
language = "English (US)",
isbn = "9780769539263",
series = "8th International Conference on Machine Learning and Applications, ICMLA 2009",
pages = "190--195",
booktitle = "8th International Conference on Machine Learning and Applications, ICMLA 2009",

}

Zhang, L, Frigui, H & Gader, P 2009, Context-dependent fusion of multiple algorithms with minimum classification error learning. in 8th International Conference on Machine Learning and Applications, ICMLA 2009., 5381831, 8th International Conference on Machine Learning and Applications, ICMLA 2009, pp. 190-195, 8th International Conference on Machine Learning and Applications, ICMLA 2009, Miami Beach, FL, United States, 12/13/09. https://doi.org/10.1109/ICMLA.2009.119

Context-dependent fusion of multiple algorithms with minimum classification error learning. / Zhang, Lijun; Frigui, Hichem; Gader, Paul.

8th International Conference on Machine Learning and Applications, ICMLA 2009. 2009. p. 190-195 5381831 (8th International Conference on Machine Learning and Applications, ICMLA 2009).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Context-dependent fusion of multiple algorithms with minimum classification error learning

AU - Zhang, Lijun

AU - Frigui, Hichem

AU - Gader, Paul

PY - 2009/12/1

Y1 - 2009/12/1

N2 - We present a novel method for fusing the decisions of multiple classification algorithms which use different features, classification methods, and data sources. The proposed method, called Context Dependent Fusion of Multiple Algorithms(CDF-MA) is motivated by the fact that the relative performance of different algorithms can vary significantly as the characteristics of the input data vary. The training part of CDF-MA has two main components: context extraction and algorithm fusion. In context extraction, the features used by the distinct algorithms are combined and clustered into groups of similar contexts. The algorithm fusion component embeds a variation of the MCE/GPD method to assigns an aggregation weight to each algorithm based on its loss function within each context. Results on real world data show that the proposed method can identify meaningful and coherent clusters, and outperform all individual classifiers and the global weighted average fusion method.

AB - We present a novel method for fusing the decisions of multiple classification algorithms which use different features, classification methods, and data sources. The proposed method, called Context Dependent Fusion of Multiple Algorithms(CDF-MA) is motivated by the fact that the relative performance of different algorithms can vary significantly as the characteristics of the input data vary. The training part of CDF-MA has two main components: context extraction and algorithm fusion. In context extraction, the features used by the distinct algorithms are combined and clustered into groups of similar contexts. The algorithm fusion component embeds a variation of the MCE/GPD method to assigns an aggregation weight to each algorithm based on its loss function within each context. Results on real world data show that the proposed method can identify meaningful and coherent clusters, and outperform all individual classifiers and the global weighted average fusion method.

UR - http://www.scopus.com/inward/record.url?scp=77950828695&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77950828695&partnerID=8YFLogxK

U2 - 10.1109/ICMLA.2009.119

DO - 10.1109/ICMLA.2009.119

M3 - Conference contribution

SN - 9780769539263

T3 - 8th International Conference on Machine Learning and Applications, ICMLA 2009

SP - 190

EP - 195

BT - 8th International Conference on Machine Learning and Applications, ICMLA 2009

ER -

Zhang L, Frigui H, Gader P. Context-dependent fusion of multiple algorithms with minimum classification error learning. In 8th International Conference on Machine Learning and Applications, ICMLA 2009. 2009. p. 190-195. 5381831. (8th International Conference on Machine Learning and Applications, ICMLA 2009). https://doi.org/10.1109/ICMLA.2009.119