Hybrid Deep Ensemble Framework for Automated Skin Cancer Detection using Advanced Optimization and Deep Learning Techniques

Abstract

This paper presents the Hybrid Deep Ensemble Framework (HDEF) for efficient and accurate skin cancer detection using the ISIC 2020 dataset. The proposed architecture integrates Convolutional Neural Networks (CNN) optimized through AdaGrad with Gated Recurrent Units (GRU) for sequential learning. Additionally, Whale Optimization Algorithm (WOA) is employed for hyperparameter tuning to enhance the model's accuracy. The HDEF model improves the diagnosis of skin cancer lesions with a focus on adaptive learning, feature extraction, and robust generalization across datasets. The results demonstrate the superiority of the HDEF framework, achieving accuracy and F1-score, outperforming conventional CNN and hybrid models