A A Study of Data Augu mentation And Its Impact on Skin Cancer Detection Using Deep Learning Techniques

Abstract

Skin cancer is a common worldwide health concern, and survival rates are greatly increased by early detection. Algorithms like Support Vector Machines (SVMs), Random Forests, and Convolutional Neural Networks (CNNs) are essential for classifying skin lesions, and machine learning (ML) has completely changed medical diagnosis. Small datasets and complex interactions are best suited for SVMs, although Random Forests provide resilience and interpretability. CNNs have revolutionized image-based diagnostics through deep learning-based hierarchical characteristic extraction. Even though these machine learning models have issues with scalability, processing costs, and dataset quality, they are incredibly accurate. CNNs outperform SVMs and Random Forests by 97%, according to experimental results. Through data augmentation, robustness is further improved, enabling ML applications in skin cancer diagnosis.