Enhancing Neural Architecture Search Using Transfer Learning and Dynamic Search Spaces for Global Horizontal Irradiance Prediction

Wedi'i Gadw mewn:
Manylion Llyfryddiaeth
Cyhoeddwyd yn:Forecasting vol. 7, no. 3 (2025), p. 43-66
Prif Awdur: Inoussa, Legrene
Awduron Eraill: Wong, Tony, Louis-A, Dessaint
Cyhoeddwyd:
MDPI AG
Pynciau:
Mynediad Ar-lein:Citation/Abstract
Full Text + Graphics
Full Text - PDF
Tagiau: Ychwanegu Tag
Dim Tagiau, Byddwch y cyntaf i dagio'r cofnod hwn!

MARC

LEADER 00000nab a2200000uu 4500
001 3254512228
003 UK-CbPIL
022 |a 2571-9394 
024 7 |a 10.3390/forecast7030043  |2 doi 
035 |a 3254512228 
045 2 |b d20250701  |b d20250930 
100 1 |a Inoussa, Legrene  |u Systems Engineering Department, École de Technologie Supérieure, Montréal, QC H3C 1K3, Canada; tony.wong@etsmtl.ca 
245 1 |a Enhancing Neural Architecture Search Using Transfer Learning and Dynamic Search Spaces for Global Horizontal Irradiance Prediction 
260 |b MDPI AG  |c 2025 
513 |a Journal Article 
520 3 |a The neural architecture search technique is used to automate the engineering of neural network models. Several studies have applied this approach, mainly in the fields of image processing and natural language processing. Its application generally requires very long computing times before converging on the optimal architecture. This study proposes a hybrid approach that combines transfer learning and dynamic search space adaptation (TL-DSS) to reduce the architecture search time. To validate this approach, Long Short-Term Memory (LSTM) models were designed using different evolutionary algorithms, including artificial bee colony (ABC), genetic algorithm (GA), differential evolution (DE), and particle swarm optimization (PSO), which were developed to predict trends in global horizontal irradiation data. The performance measures of this approach include the performance of the proposed models, as evaluated via RMSE over a 24-h prediction window of the solar irradiance data trend on one hand, and CPU search time on the other. The results show that, in addition to reducing the search time by up to 89.09% depending on the search algorithm, the proposed approach enables the creation of models that are up to 99% more accurate than the non-enhanced approach. This study demonstrates that it is possible to reduce the search time of a neural architecture while ensuring that models achieve good performance. 
653 |a Particle swarm optimization 
653 |a Evolutionary computation 
653 |a Swarm intelligence 
653 |a Datasets 
653 |a Deep learning 
653 |a Convergence 
653 |a Learning curves 
653 |a Solar energy 
653 |a Neural networks 
653 |a Genetic algorithms 
653 |a Computer architecture 
653 |a Optimization 
653 |a Renewable resources 
653 |a Search algorithms 
653 |a Algorithms 
653 |a Alternative energy sources 
653 |a Machine learning 
653 |a Natural language processing 
653 |a Image processing 
653 |a Irradiance 
653 |a Radiation 
653 |a Evolutionary algorithms 
700 1 |a Wong, Tony  |u Systems Engineering Department, École de Technologie Supérieure, Montréal, QC H3C 1K3, Canada; tony.wong@etsmtl.ca 
700 1 |a Louis-A, Dessaint  |u Electrical Engineering Department, École de Technologie Supérieure, Montréal, QC H3C 1K3, Canada; louis-a.dessaint@etsmtl.ca 
773 0 |t Forecasting  |g vol. 7, no. 3 (2025), p. 43-66 
786 0 |d ProQuest  |t ABI/INFORM Global 
856 4 1 |3 Citation/Abstract  |u https://www.proquest.com/docview/3254512228/abstract/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch 
856 4 0 |3 Full Text + Graphics  |u https://www.proquest.com/docview/3254512228/fulltextwithgraphics/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch 
856 4 0 |3 Full Text - PDF  |u https://www.proquest.com/docview/3254512228/fulltextPDF/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch