Paper Title
Skin Cancer Classification and Disease Detection using Light Weight and Deep Learning Techniques

Abstract
Skin cancer is one of the most common and potentially life-threatening forms of cancer that affect people around the world. However, if recognized early stages, it can be handled effectively and significantly improves patient outcomes. While traditional diagnostic methods are effective, they are often labor intensive, time-consuming and require considerable computational resources. This is not suitable for real-time or large scale screening, especially in distant or resource-limited environments. Dermatological images from HAM10000 data records are used to classify different types of skin lesions. The proposed system uses techniques such as data augmentation to increase data set variability, send learning and use educated knowledge, and finetune the model architecture to optimize performance for a particular task. These models are known for their reduced number of parameters and faster inference times. This means it's ideal for providing device for resourcerelated devices such as smartphones, Raspberry PIs and other embedded systems. This allows for real-time skin analysis of mobile health streets, which can support dermatologists and non-specialists with early detection. Integrating optical models into diagnostic tools could revolutionize matology screening by becoming sustainable, scalable, and inexpensive, expanding access to healthcare groups to adequate population groups. Keywords - Lightweight Deep Learning,HAM10000 Dataset, Dermatoscopic Images