Benchmarking of deep learning algorithms for skin cancer detection based on a hybrid framework of entropy and VIKOR techniques

Skin cancer is one of the most common cancers worldwide caused by excessive development skin cells. Considering rapid growth use deep learning algorithms for detection, selecting optimal algorithm has become crucial to determining efficiency computer-aided diagnosis (CAD) systems developed healthcare sector. However, a sufficient number criteria and parameters must be considered when an ideal algorithm. A generally accepted method benchmarking models classification unavailable in current literature. This paper presents multi-criteria decision-making framework evaluating detection based on hybridisation entropy VIseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) methods. Twelve well-known convolution networks are evaluated tested eleven publicly available image datasets achieve target study. Several related convolutional neural (CNNs) architectures, including optimisation technique, transfer learning, class balancing, data augmentation, network complexity, have been evaluation. The decision matrix (DM) designed crossover five evaluation twelve different datasets. Subsequently, ranking models, making (MCDM) techniques used. MCDM uses scheme that involves integration with VIKOR approaches. For weight calculations criteria, applied, while used benchmark rank models. obtained results reveal InceptionResNetV2 model gained first selected as architecture considering investigated our presented achieves significant performance best algorithm, which could provide substantial guidance researcher working field.