Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm
In recent years, Convolutional Neural Networks (CNNs) have emerged as powerful tools for solving complex real-world problems, particularly in the domain of image processing. The success of CNNs can be attributed to their ability to learn hierarchical representations from data. However, achieving opt...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11087552/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849687443122421760 |
|---|---|
| author | Abdel-Hamid M. Emara Ghada Atteia Jawad Hasan Alkhateeb |
| author_facet | Abdel-Hamid M. Emara Ghada Atteia Jawad Hasan Alkhateeb |
| author_sort | Abdel-Hamid M. Emara |
| collection | DOAJ |
| description | In recent years, Convolutional Neural Networks (CNNs) have emerged as powerful tools for solving complex real-world problems, particularly in the domain of image processing. The success of CNNs can be attributed to their ability to learn hierarchical representations from data. However, achieving optimal performance with CNNs often necessitates fine-tuning a myriad of hyperparameters, such as learning rates, batch sizes, and network architectures. This tuning process typically relies on expert judgment and can be time-consuming and resource-intensive. To address this challenge, a novel metaheuristic-based optimization approach called Accelerated Particle Swarm Optimization for CNNs (APSO-CNN) is proposed. APSO-CNN uses the global search capabilities of Particle Swarm Optimization (PSO) to automatically optimize hyperparameter configurations for architecture-determined Convolutional Neural Networks (CNNs). The main contribution of this work is the development of the APSO-CNN framework, which introduces an improved PSO variant specifically tailored for Convolutional Neural Networks (CNNs) hyperparameter tuning, thereby reducing manual intervention and significantly enhancing model performance. The proposed APSO-CNN is evaluated across various Convolutional Neural Networks (CNNs) architectures and the Faces94 dataset, representing a diverse range of image processing tasks. Experiments demonstrate that the proposed automated hyperparameter fine-tuning approach consistently yields significant improvements in performance metrics, including Accuracy, Precision, Recall, and F1 Score. The APSO-CNN framework was experimentally validated on the Faces94 dataset comprising 3,060 facial images. These enhancements underscore the effectiveness of APSO-CNN in optimizing Convolutional Neural Networks (CNNs) for image processing applications. Accelerated Particle Swarm Optimization for CNNs (APSO-CNN) presents a promising solution to the challenge of hyperparameter optimization in Convolutional Neural Networks (CNNs). By automating this critical aspect of model development, Accelerated Particle Swarm Optimization for CNNs (APSO-CNN), streamlines the process, making it more efficient and accessible to researchers and practitioners. This contribution has the potential to accelerate advancements in image processing and related fields by enabling the rapid development of high-performance Convolutional Neural Networks (CNNs) models. |
| format | Article |
| id | doaj-art-5163bc4bbe514c59bd56503c32e059c4 |
| institution | DOAJ |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-5163bc4bbe514c59bd56503c32e059c42025-08-20T03:22:19ZengIEEEIEEE Access2169-35362025-01-011313450613451810.1109/ACCESS.2025.359140311087552Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization AlgorithmAbdel-Hamid M. Emara0Ghada Atteia1https://orcid.org/0000-0002-5462-595XJawad Hasan Alkhateeb2https://orcid.org/0000-0001-7611-7887Department of Computers and Systems Engineering, Faculty of Engineering, Al-Azhar University, Cairo, EgyptDepartment of Information Technology, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, P.O. Box 84428, Riyadh, Saudi ArabiaComputer Engineering Department, College of Computer Engineering and Science, Prince Mohammad Bin Fahd University, Al-Khobar, Saudi ArabiaIn recent years, Convolutional Neural Networks (CNNs) have emerged as powerful tools for solving complex real-world problems, particularly in the domain of image processing. The success of CNNs can be attributed to their ability to learn hierarchical representations from data. However, achieving optimal performance with CNNs often necessitates fine-tuning a myriad of hyperparameters, such as learning rates, batch sizes, and network architectures. This tuning process typically relies on expert judgment and can be time-consuming and resource-intensive. To address this challenge, a novel metaheuristic-based optimization approach called Accelerated Particle Swarm Optimization for CNNs (APSO-CNN) is proposed. APSO-CNN uses the global search capabilities of Particle Swarm Optimization (PSO) to automatically optimize hyperparameter configurations for architecture-determined Convolutional Neural Networks (CNNs). The main contribution of this work is the development of the APSO-CNN framework, which introduces an improved PSO variant specifically tailored for Convolutional Neural Networks (CNNs) hyperparameter tuning, thereby reducing manual intervention and significantly enhancing model performance. The proposed APSO-CNN is evaluated across various Convolutional Neural Networks (CNNs) architectures and the Faces94 dataset, representing a diverse range of image processing tasks. Experiments demonstrate that the proposed automated hyperparameter fine-tuning approach consistently yields significant improvements in performance metrics, including Accuracy, Precision, Recall, and F1 Score. The APSO-CNN framework was experimentally validated on the Faces94 dataset comprising 3,060 facial images. These enhancements underscore the effectiveness of APSO-CNN in optimizing Convolutional Neural Networks (CNNs) for image processing applications. Accelerated Particle Swarm Optimization for CNNs (APSO-CNN) presents a promising solution to the challenge of hyperparameter optimization in Convolutional Neural Networks (CNNs). By automating this critical aspect of model development, Accelerated Particle Swarm Optimization for CNNs (APSO-CNN), streamlines the process, making it more efficient and accessible to researchers and practitioners. This contribution has the potential to accelerate advancements in image processing and related fields by enabling the rapid development of high-performance Convolutional Neural Networks (CNNs) models.https://ieeexplore.ieee.org/document/11087552/Automated hyperparameter tuningconvolutional neural networks (CNNs)hyperparameter optimizationimage processingparticle swarm optimization (PSO)recognition accuracy |
| spellingShingle | Abdel-Hamid M. Emara Ghada Atteia Jawad Hasan Alkhateeb Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm IEEE Access Automated hyperparameter tuning convolutional neural networks (CNNs) hyperparameter optimization image processing particle swarm optimization (PSO) recognition accuracy |
| title | Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm |
| title_full | Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm |
| title_fullStr | Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm |
| title_full_unstemmed | Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm |
| title_short | Fine Tuning Hyperparameters of Deep Learning Models Using Metaheuristic Accelerated Particle Swarm Optimization Algorithm |
| title_sort | fine tuning hyperparameters of deep learning models using metaheuristic accelerated particle swarm optimization algorithm |
| topic | Automated hyperparameter tuning convolutional neural networks (CNNs) hyperparameter optimization image processing particle swarm optimization (PSO) recognition accuracy |
| url | https://ieeexplore.ieee.org/document/11087552/ |
| work_keys_str_mv | AT abdelhamidmemara finetuninghyperparametersofdeeplearningmodelsusingmetaheuristicacceleratedparticleswarmoptimizationalgorithm AT ghadaatteia finetuninghyperparametersofdeeplearningmodelsusingmetaheuristicacceleratedparticleswarmoptimizationalgorithm AT jawadhasanalkhateeb finetuninghyperparametersofdeeplearningmodelsusingmetaheuristicacceleratedparticleswarmoptimizationalgorithm |