Melanoma is a major skin cancer type that has a very high death rate. The various sorts of skin abrasions cause an imprecise analysis because of their high resemblance. Precise categorization of the skin abrasions in the premature phase will allow dermatologists to cure the affected individuals in well time and hence saving their lives. This is backed by a research that shows that 90% of the cases are curable, if identified in the initial phase. With the advancements in the computing power and image classification, automatic detection of the melanoma using computer algorithms has become far reliable. With many methods used, neural networks prove to be the best solution devised to attain the highest accuracy in classifying melanoma through early symptoms. We did our survey to find the drawbacks of recent models that serve this purpose with the goal to overcome them and provide a better solution. With the findings based on this survey, we proposed a model that stives to overcome the drawbacks concluded from the previous models. With an accuracy of ~96%, the proposed model provides better solution in prediciting whether the skin lesions are malignant or not.