Skin Cancer Detection Using Ensemble Learning: ViT, InceptionV3, and MobileNetV2

Skin cancer is one of the most common types of cancer globally, and early detection plays a crucial role in improving patient survival rates. However, automatic classification of skin cancer still faces significant challenges due to the diversity of colors and textures in skin lesions. This study aims to enhance accuracy and performance in minimizing false negatives in skin cancer detection by proposing a deep learning-based framework that combines the strengths of Vision Transformer (ViT), InceptionV3, and MobileNetV2 within an ensemble model. The dataset used to train and evaluate the model is ISIC 2024. Experimental results show that the ensemble learning approach achieves a classification accuracy of 99.45% and improves performance in reducing false negatives in skin cancer detection. By leveraging the advantages of each architecture, the ensemble model overcomes the individual limitations of each model. Thus, this study makes a significant contribution to the development of more reliable skin cancer detection systems to support automated diagnosis, with the potential for practical implementation in clinical applications.