MELANOMA SKIN LESION CLASSIFICATION USING IMPROVED EFFICIENTNETB3 10.5455/jjcit.71-1636005929 Saumya R. Salian,Sudhir D. Sawarkar Melanoma,Deep learning,Classification,Skin lesions,Computer vision 596 143 4-Nov.-2021 26-Dec.-2021 12-Jan.-2022 Malignant skin cancer is one the most common and lethal type of skin cancer. Early detection of cancerous skin lesions will increase the possibility of patient survival. In recent years, implementation of models built on deep neural networks in building medical diagnostic imaging systems is quite beneficial to medical experts. In this study, we present an improved and fine-tuned EfficientNetB3 model to classify malignant skin lesions using the concept of fine-tuning transfer learning. We have performed a comparative analysis of different deep learning pre-trained models, like ResNet50, InceptionV3, InceptionResNetV2 and EfficientNet B0-B2 models. The analysis findings signify the ability of utilizing fine-tuned EfficientNetB3 in the mission of melanoma detection and development of a computer-aided diagnostic system. All experimental procedures were carried out on ISBI-ISIC 2017 dataset. To check the efficiency of the proposed model, we compare the proposed model with EfficientNetB3 baseline model and present state-of-art pre-trained methods and approaches. The proposed EfficientNetB3 model obtained an accuracy of 87.12%, a recall of 87.00%, a precision of 87.00% and an F1 score of 85.00%. The proposed model achieved good computational results and efficaciously addressed the problem of model over- fitting and abated false negative labels.