Human Skin Disease Detection using Convolutional Neural Network Method with Hyperparameter Tuning to Determine the Best Parameter Combination
Riki Martua Aritonang, Mangaras Yanu Florestiyanto, Bambang Yuwono · Telematika · 2023
Purpose: Obtaining the best hyperparameter combination for optimization of the Convolutional Neural Network method, for classifying skin diseases.Design/methodology/approach: Using the CNN method with hyperparameter tuning in determining the best hyperparameter combination. System development is performed with the Python programming language.Findings/result: The best combination of hyperparameter tuning results is RMSprop optimizer, APL dropout value is 0.05, dropout is 0.5 , dense layer is 64, and produces an accuracy of 97,81%.Originality/value/state of the art: This study has differences in terms of the types of skin diseases classified, the architecture of the CNN model, the hyperparameters tested and the combination results obtained compared to previous studies.