Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks

The paper presents the GeNNsem (Genetic algorithm ANNs ensemble) software framework for the simultaneous optimization of individual neural networks and building their optimal ensemble. The proposed framework employs a genetic algorithm to search for suitable architectures and hyperparameters of the...

Full description

Saved in:
Bibliographic Details
Main Authors: Lazar Krstic, Milos Ivanovic, Visnja Simic, Boban Stojanovic
Format: Article
Language:English
Published: Elsevier 2024-12-01
Series:Egyptian Informatics Journal
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1110866524001440
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850119400834727936
author Lazar Krstic
Milos Ivanovic
Visnja Simic
Boban Stojanovic
author_facet Lazar Krstic
Milos Ivanovic
Visnja Simic
Boban Stojanovic
author_sort Lazar Krstic
collection DOAJ
description The paper presents the GeNNsem (Genetic algorithm ANNs ensemble) software framework for the simultaneous optimization of individual neural networks and building their optimal ensemble. The proposed framework employs a genetic algorithm to search for suitable architectures and hyperparameters of the individual neural networks to maximize the weighted sum of accuracy and diversity in their predictions. The optimal ensemble consists of networks with low errors but diverse predictions, resulting in a more generalized model. The scalability of the proposed framework is ensured by utilizing micro-services and Kubernetes batching orchestration. GeNNsem has been evaluated on two regression benchmark problems and compared with related machine learning techniques. The proposed approach exhibited supremacy over other ensemble approaches and individual neural networks in all common regression modeling metrics. Real-world use-case experiments in the domain of hydro-informatics have further demonstrated the main advantages of GeNNsem: requires the least training sessions for individual models when optimizing an ensemble; networks in an ensemble are generally simple due to the regularization provided by a trivial initial population and custom genetic operators; execution times are reduced by two orders of magnitude as a result of parallelization.
format Article
id doaj-art-4b73c628aa5d4e8a84a1d84a41bf6a07
institution OA Journals
issn 1110-8665
language English
publishDate 2024-12-01
publisher Elsevier
record_format Article
series Egyptian Informatics Journal
spelling doaj-art-4b73c628aa5d4e8a84a1d84a41bf6a072025-08-20T02:35:39ZengElsevierEgyptian Informatics Journal1110-86652024-12-012810058110.1016/j.eij.2024.100581Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networksLazar Krstic0Milos Ivanovic1Visnja Simic2Boban Stojanovic3Corresponding author.; University of Kragujevac, Faculty of Science, 12 Radoja Domanovica Street, 34000 Kragujevac, SerbiaUniversity of Kragujevac, Faculty of Science, 12 Radoja Domanovica Street, 34000 Kragujevac, SerbiaUniversity of Kragujevac, Faculty of Science, 12 Radoja Domanovica Street, 34000 Kragujevac, SerbiaUniversity of Kragujevac, Faculty of Science, 12 Radoja Domanovica Street, 34000 Kragujevac, SerbiaThe paper presents the GeNNsem (Genetic algorithm ANNs ensemble) software framework for the simultaneous optimization of individual neural networks and building their optimal ensemble. The proposed framework employs a genetic algorithm to search for suitable architectures and hyperparameters of the individual neural networks to maximize the weighted sum of accuracy and diversity in their predictions. The optimal ensemble consists of networks with low errors but diverse predictions, resulting in a more generalized model. The scalability of the proposed framework is ensured by utilizing micro-services and Kubernetes batching orchestration. GeNNsem has been evaluated on two regression benchmark problems and compared with related machine learning techniques. The proposed approach exhibited supremacy over other ensemble approaches and individual neural networks in all common regression modeling metrics. Real-world use-case experiments in the domain of hydro-informatics have further demonstrated the main advantages of GeNNsem: requires the least training sessions for individual models when optimizing an ensemble; networks in an ensemble are generally simple due to the regularization provided by a trivial initial population and custom genetic operators; execution times are reduced by two orders of magnitude as a result of parallelization.http://www.sciencedirect.com/science/article/pii/S1110866524001440Ensemble modelingRegressionANNGenetic algorithmDistributed computing
spellingShingle Lazar Krstic
Milos Ivanovic
Visnja Simic
Boban Stojanovic
Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
Egyptian Informatics Journal
Ensemble modeling
Regression
ANN
Genetic algorithm
Distributed computing
title Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
title_full Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
title_fullStr Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
title_full_unstemmed Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
title_short Evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
title_sort evolutionary approach for composing a thoroughly optimized ensemble of regression neural networks
topic Ensemble modeling
Regression
ANN
Genetic algorithm
Distributed computing
url http://www.sciencedirect.com/science/article/pii/S1110866524001440
work_keys_str_mv AT lazarkrstic evolutionaryapproachforcomposingathoroughlyoptimizedensembleofregressionneuralnetworks
AT milosivanovic evolutionaryapproachforcomposingathoroughlyoptimizedensembleofregressionneuralnetworks
AT visnjasimic evolutionaryapproachforcomposingathoroughlyoptimizedensembleofregressionneuralnetworks
AT bobanstojanovic evolutionaryapproachforcomposingathoroughlyoptimizedensembleofregressionneuralnetworks