Melanoma Skin Cancer Detection Using Ensemble of Machine Learning Models with ResNeXt101 and TinyViT

Abstract

Although skin cancer is a deadly condition, it can be effectively treated with a quick test. Still, it might be challenging to diagnose skin conditions in a timely manner. While most of the earlier work may be covered by a single model system, integrating multiple models helps increase classification accuracy. Early research usually used Deep Convolutional Neural Networks (DCNN), which struggle to capture global properties. ResNeXt101 and TinyViT are the two learning models that make up this hybrid model. The model performs better overall as a result of this integration. More significantly, when applied to melanoma cancer data, the hybrid model performs as a weighted model and achieves an astounding 91.6% accuracy.