Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

5 Citaties (Scopus)
48 Downloads (Pure)

Uittreksel

Many real-world control and classification tasks involve a large number of features. When artificial neural networks (ANNs) are used for modeling these tasks, the network architectures tend to be large. Neuroevolution is an effective approach for optimizing ANNs; however, there are two bottlenecks that make their application challenging in case of high-dimensional networks using direct encoding. First, classic evolutionary algorithms tend not to scale well for searching large parameter spaces; second, the network evaluation over a large number of training instances is in general time-consuming. In this work, we propose an approach called the Limited Evaluation Cooperative Co-evolutionary Differential Evolution algorithm (LECCDE) to optimize high-dimensional ANNs. The proposed method aims to optimize the pre-synaptic weights of each post-synaptic neuron in different subpopulations using a Cooperative Co-evolutionary Differential Evolution algorithm, and employs a limited evaluation scheme where fitness evaluation is performed on a relatively small number of training instances based on fitness inheritance. We test LECCDE on three datasets with various sizes, and our results show that cooperative co-evolution significantly improves the test error comparing to standard Differential Evolution, while the limited evaluation scheme facilitates a significant reduction in computing time.

Originele taal-2Engels
TitelGECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference
UitgeverijAssociation for Computing Machinery, Inc
Pagina's569-576
Aantal pagina's8
ISBN van elektronische versie9781450356183
DOI's
StatusGepubliceerd - 2 jul 2018
Evenement2018 Genetic and Evolutionary Computation Conference, GECCO 2018 - Kyoto, Japan
Duur: 15 jul 201819 jul 2018

Congres

Congres2018 Genetic and Evolutionary Computation Conference, GECCO 2018
LandJapan
StadKyoto
Periode15/07/1819/07/18

Vingerafdruk

Neural networks
Network architecture
Evolutionary algorithms
Neurons

Citeer dit

Yaman, A., Mocanu, D. C., Iacca, G., Fletcher, G., & Pechenizkiy, M. (2018). Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution. In GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference (blz. 569-576). Association for Computing Machinery, Inc. https://doi.org/10.1145/3205455.3205555
Yaman, Anil ; Mocanu, Decebal Constantin ; Iacca, Giovanni ; Fletcher, George ; Pechenizkiy, Mykola. / Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution. GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference. Association for Computing Machinery, Inc, 2018. blz. 569-576
@inproceedings{25ce5e7a7dcd4327883237e6cf13c12a,
title = "Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution",
abstract = "Many real-world control and classification tasks involve a large number of features. When artificial neural networks (ANNs) are used for modeling these tasks, the network architectures tend to be large. Neuroevolution is an effective approach for optimizing ANNs; however, there are two bottlenecks that make their application challenging in case of high-dimensional networks using direct encoding. First, classic evolutionary algorithms tend not to scale well for searching large parameter spaces; second, the network evaluation over a large number of training instances is in general time-consuming. In this work, we propose an approach called the Limited Evaluation Cooperative Co-evolutionary Differential Evolution algorithm (LECCDE) to optimize high-dimensional ANNs. The proposed method aims to optimize the pre-synaptic weights of each post-synaptic neuron in different subpopulations using a Cooperative Co-evolutionary Differential Evolution algorithm, and employs a limited evaluation scheme where fitness evaluation is performed on a relatively small number of training instances based on fitness inheritance. We test LECCDE on three datasets with various sizes, and our results show that cooperative co-evolution significantly improves the test error comparing to standard Differential Evolution, while the limited evaluation scheme facilitates a significant reduction in computing time.",
keywords = "Cooperative Co-evolution, Differential evolution, Direct encoding, Neuroevolution",
author = "Anil Yaman and Mocanu, {Decebal Constantin} and Giovanni Iacca and George Fletcher and Mykola Pechenizkiy",
year = "2018",
month = "7",
day = "2",
doi = "10.1145/3205455.3205555",
language = "English",
pages = "569--576",
booktitle = "GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference",
publisher = "Association for Computing Machinery, Inc",
address = "United States",

}

Yaman, A, Mocanu, DC, Iacca, G, Fletcher, G & Pechenizkiy, M 2018, Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution. in GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference. Association for Computing Machinery, Inc, blz. 569-576, Kyoto, Japan, 15/07/18. https://doi.org/10.1145/3205455.3205555

Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution. / Yaman, Anil; Mocanu, Decebal Constantin; Iacca, Giovanni; Fletcher, George; Pechenizkiy, Mykola.

GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference. Association for Computing Machinery, Inc, 2018. blz. 569-576.

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

TY - GEN

T1 - Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution

AU - Yaman, Anil

AU - Mocanu, Decebal Constantin

AU - Iacca, Giovanni

AU - Fletcher, George

AU - Pechenizkiy, Mykola

PY - 2018/7/2

Y1 - 2018/7/2

N2 - Many real-world control and classification tasks involve a large number of features. When artificial neural networks (ANNs) are used for modeling these tasks, the network architectures tend to be large. Neuroevolution is an effective approach for optimizing ANNs; however, there are two bottlenecks that make their application challenging in case of high-dimensional networks using direct encoding. First, classic evolutionary algorithms tend not to scale well for searching large parameter spaces; second, the network evaluation over a large number of training instances is in general time-consuming. In this work, we propose an approach called the Limited Evaluation Cooperative Co-evolutionary Differential Evolution algorithm (LECCDE) to optimize high-dimensional ANNs. The proposed method aims to optimize the pre-synaptic weights of each post-synaptic neuron in different subpopulations using a Cooperative Co-evolutionary Differential Evolution algorithm, and employs a limited evaluation scheme where fitness evaluation is performed on a relatively small number of training instances based on fitness inheritance. We test LECCDE on three datasets with various sizes, and our results show that cooperative co-evolution significantly improves the test error comparing to standard Differential Evolution, while the limited evaluation scheme facilitates a significant reduction in computing time.

AB - Many real-world control and classification tasks involve a large number of features. When artificial neural networks (ANNs) are used for modeling these tasks, the network architectures tend to be large. Neuroevolution is an effective approach for optimizing ANNs; however, there are two bottlenecks that make their application challenging in case of high-dimensional networks using direct encoding. First, classic evolutionary algorithms tend not to scale well for searching large parameter spaces; second, the network evaluation over a large number of training instances is in general time-consuming. In this work, we propose an approach called the Limited Evaluation Cooperative Co-evolutionary Differential Evolution algorithm (LECCDE) to optimize high-dimensional ANNs. The proposed method aims to optimize the pre-synaptic weights of each post-synaptic neuron in different subpopulations using a Cooperative Co-evolutionary Differential Evolution algorithm, and employs a limited evaluation scheme where fitness evaluation is performed on a relatively small number of training instances based on fitness inheritance. We test LECCDE on three datasets with various sizes, and our results show that cooperative co-evolution significantly improves the test error comparing to standard Differential Evolution, while the limited evaluation scheme facilitates a significant reduction in computing time.

KW - Cooperative Co-evolution

KW - Differential evolution

KW - Direct encoding

KW - Neuroevolution

UR - http://www.scopus.com/inward/record.url?scp=85050642203&partnerID=8YFLogxK

U2 - 10.1145/3205455.3205555

DO - 10.1145/3205455.3205555

M3 - Conference contribution

AN - SCOPUS:85050642203

SP - 569

EP - 576

BT - GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference

PB - Association for Computing Machinery, Inc

ER -

Yaman A, Mocanu DC, Iacca G, Fletcher G, Pechenizkiy M. Limited evaluation cooperative co-evolutionary differential evolution for large-scale neuroevolution. In GECCO 2018 - Proceedings of the 2018 Genetic and Evolutionary Computation Conference. Association for Computing Machinery, Inc. 2018. blz. 569-576 https://doi.org/10.1145/3205455.3205555