In the current scenario, the death rate due to the cause of skin cancer is increasing enormously. Diagnosis and prediction of Skin Cancer (SC) have become vital at an earlier stage. The main objective of this research is ensemble machine learning with enhanced genetic algorithm technique to achieve higher accuracy in the prediction of skin cancer at an earlier stage compared to other existing techniques. Although many machine learning and deep learning approaches implemented in detecting skin cancer at an earlier stage still there are few limitations. To overcome these problems in our proposed work, the CNN model, ResNet-16 usually produces successful results in extracting the features automatically and classifying the images very accurately. Therefore, the ResNet model used in our work obtains the deep features with the help of a fully connected layer. Later the feature selection is performed with the help of an Enhanced Genetic Algorithm (EGA) that produces optimized solutions by implementing operations like mutations, crossover, and ensemble with Extreme Learning Machine (EGA-ELM) to classify the images as either melanoma or non-melanoma. The proposed model certainly achieved higher accuracy and effective performance. Finally, the obtained results are to be compared with other popular classifying algorithms like Support Vector Machine (SVM) and various other models.

Skin Cancer Genetic Algorithm Extreme Learning Machine (ELM) CNN Optimization