Review of Melanoma Detection by using Blending Ensemble Method

The most severe kind of skin cancer in humans is melanoma. Due to melanoma detection delays, most humans die for years. Effective treatment of melanoma skin cancer depends on early detection. The automatic system will decrease the number of fatal instances brought on by melanoma. In the pathology lab, there are some manual methods for finding melanoma. However, this method may take more time because of the complex dataset. Deep learning models are used to solve this issue. There are numerous melanoma datasets on Kaggle. This experiment will use the ISIC dataset downloaded from Kaggle. The primary goals of the proposed approach are to decrease overfitting, improve model accuracy, minimize error, shorten training periods, and address issues with class imbalance. In a different research paper study, we observed that the ensemble approach in deep learning can give better accuracy than when individual models are used. For the experiment, use blending ensemble learning. This project will use a base model such as CNN, Vgg-16, GoogleNet, and ResNet-50 as a meta-model for the final prediction. The model evaluation will use the confusion matrix, accuracy, and F1 score.