Enhancing Accuracy for Classification using the CNN Model and Hyperparameter Optimization Algorithm

Abstract

The Convolutional Neural Network (CNN) is one of the most widely used deep learning models, particularly effective for image recognition and classification tasks. The accuracy of a CNN depends not only on its network structure but also critically on its hyperparameters. Therefore, identifying optimal hyperparameters is key to enhancing the CNN model's performance. In this study, the authors propose using optimization algorithms such as Random Search, Bayesian Optimization with Gaussian processes, and Bayesian Optimization with Tree-structured Parzen Estimators to optimize the hyperparameters of the CNN model. The results are compared with those from traditional machine learning models, including Random Forest, Support Vector Classification, and K-Nearest NeigHPOr. Both the MNIST and Olivetti Faces datasets are used at the same time in this research. The findings demonstrate that integrating these optimization algorithms with the CNN model improves prediction accuracy over traditional models during both training and the train-test phases.