<![CDATA[Image-based malware detection has become an area of further research in dealing with image-based malware attacks. Various deep learning models have been used to improve detection accuracy. One popular architecture is VGG16, a convolutional network widely used in image classification. In this study, we explore the impact of hyperparameter tuning on the optimization of the VGG16 model for image-based malware detection. The hyperparameter experiments conducted in this study are optimizer, and the number of epochs. Through 6 experiments with parameter variations, we evaluate the performance of the VGG16 model using several SGD, and Adam optimizers and the number of epochs consisting of 100, 250 and 500 epochs. The experimental results show that the selection and tuning of the optimizer can affect the performance of the model in terms of accuracy and training efficiency. The optimized Adam optimizer gives the best results, with higher detection accuracy than the SGD optimizer. The results show that the Adam optimizer has the highest accuracy reaching 85%.]]>
Copyrights © 2024
