Optimizing Convolutional Neural Network Hyperparameters Using the Secretary Bird Optimization Algorithm

کد مقاله : 1175-NAEC

نویسندگان

ساره گرگ بندی * ندارم

چکیده مقاله

Convolutional Neural Networks (CNN) are one of the most common topics in deep learning (DL) research due to their architectural advantages. CNN are highly dependent on hyperparameter configurations, and manually tuning these hyperparameters can be time-consuming for researchers. Therefore, efficient optimization techniques are needed. Metaheuristic optimization algorithms can be used to determine the optimal hyperparameters of CNN. In this study, the Secretary Bird Optimization Algorithm (SBOA) is used to optimize the hyperparameters of CNN, such as the number of filters in convolutional layers, the size of convolutional filters, Learning Rate, and Batch Size. This design was applied to the MNIST handwritten digit dataset, and the model's performance result shows an accuracy of 99.6%. dependent on hyperparameter configurations, and manually tuning these hyperparameters can be time-consuming for researchers. Therefore, efficient optimization techniques are neededy dependent on hyperparameter configurations, and manually tuning these hyperparameters can be time-consuming for researchers. Therefore, efficient optimization techniques are needed. Metaheuristic optimization algorithms can be used to determine the optimal hyperparameters of CNN. In this study, the Secretary Bird Optimization Algorithm (SBOA) is used to optimize the hyperparameters of CNN, such as the number of filters in convolutional layers, the size of convolutional filters, Learning Rate, and Batch Size. This design was applied to the MNIST handwritten digit dataset, and the model's performance result shows an accuracy of 99.6%. dependent on hyperparameter configurations, and manually tuning these hyperparameters can be time-consuming for researchers. Therefore, efficient optimization techniques are needed

کلیدواژه ها

Keywords: Convolutional Neural Networks (CNN), Deep Learning (DL), Hyperparameter, Metaheuristic Algorithms, Optimization; SBOA

وضعیت: پذیرفته شده