<![CDATA[Skin diseases are often caused by air temperature, environmental hygiene and personal hygiene, with symptoms such as itching, pain and redness. Contributing factors include exposure to chemicals, sunlight, infections, a weak immune system, microorganisms, and poor personal hygiene. This study uses Convolutional Neural Networks (CNN) with EfficientNetB0 model and hyperparameter optimization for skin disease classification. The dataset consists of 1158 images that have been divided into eight categories, with 80% for training data and 20% for test data. Data augmentation is applied to increase the variety of training data. Various combinations of hyperparameters such as learning rate, optimizer (Adamax and AdamW), and activation function (ReLU and LeakyReLU) were tested in 16 training scenarios. The best results was obtained from the third scenario with the original dataset, Adamax optimizer, ReLU activation function, and 0.01 learning rate, which gave a testing accuracy of 95.70%. The model also showed good generalization and low loss values. Confusion matrix analysis and classification report showed high accuracy in skin disease classification. This study concludes that EfficientNetB0 with proper hyperparameter optimization can improve the accuracy and effectiveness of skin disease diagnosis.]]>
Copyrights © 2025
