Trained neural networking framework based skin cancer diagnosis and categorization using grey wolf optimization.

Area of Science:

• Dermatology and Artificial Intelligence
• Computational Biology and Bioinformatics

Background:

• Skin cancer diagnosis relies on analyzing epidermal mutations and visual characteristics.
• Current diagnostic methods often use independent datasets, limiting generalizability and real-world applicability.
• Developing robust and accurate skin cancer categorization techniques is crucial for early detection and treatment.

Purpose of the Study:

• To optimize and categorize Kaggle-based skin cancer datasets using Federated Learning (FL).
• To extract dataset attribute dependencies and perform dimensional mapping for improved analysis.
• To validate and train the optimized datasets within a neural networking framework enhanced by FL.

Main Methods:

• Federated Learning (FL) was employed for dataset optimization and categorization.
• Grey Wolf Optimization (GWO) algorithm was utilized for extracting dataset attribute dependencies and dimensional mapping.
• A neural networking framework, expanded by FL standards, was used for threshold value validation and training.

Main Results:

• The GWO technique demonstrated a high accuracy of 95.82% in the categorization task.
• The combined approach of Trained Neural Networking (TNN) and Recessive Learning (RL) achieved 94.9% accuracy.
• The methodology effectively processed and indexed labeled data arrays from optimized datasets.

Conclusions:

• The proposed FL-based approach, integrated with GWO, significantly improves skin cancer dataset categorization and diagnostic accuracy.
• The study highlights the potential of federated learning standards in enhancing neural network performance for medical image analysis.
• Accurate data processing and attribute extraction are key to developing reliable AI-driven diagnostic tools for skin cancer.