Paper Title
Skin Lesion Classification and Prediction - A Comparative Analysis of Machine Learning Techniques

Abstract
One of the deadliest types of skin cancer, melanoma, needs to be diagnosed as soon as possible in order to increase patient survival rates.This study explored the use of machine learning for the automated categorization of cancerous lesions in dermoscopic images. Models like Convolutional Neural Networks (CNNs), ResNet50, InceptionV3 and DenseNet121 were tested on the HAM10000 benchmark dataset with and without data augmentation.The primary measures of performance were accuracy, precision, recall, and F1 score. Out of all the models evaluated, the CNN model achieved the highest accuracy of 97% with the application of data augmentation. In general, data augmentation proved to greatly improve the accuracy of the model. Furthermore, efforts were made to further enhance performance by implementing techniques such as linear discriminant analysis (LDA) and principal component analysis (PCA) to reduce dimensionality. While PCA decreased accuracy for some models, LDA improved it in select cases. Results demonstrate machine learning, especially CNNs, can accurately distinguish benign and malignant skin lesions. Automating this screening could help dermatologists reduce late-stage diagnosis. Widespread adoption of these techniques may lower treatment costs, save lives, and pave the way for automated diagnosis in other medical imaging areas Keywords - Machine Learning (ML), Convolutional Neural Networks(Cnns), Skin Cancer, Dermoscopic Images, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA)