Taylor‐Based Adaptive Neuro‐Fuzzy‐Shepard <scp>CNN</scp> for Skin Cancer Classification

ABSTRACT Skin cancer is a deadly illness that spreads quickly around the world. As the lesions in the skin are similar, detecting skin cancer at an initial stage is a difficult process. Hence, to solve this issue, a new method is devised for classifying skin cancer, and it is called a Taylor‐based adaptive neuro‐fuzzy‐Shepard convolutional neural network (TANF‐ShCNN). First, an input skin image is preprocessed by a bilateral filter for eradicating the noise. Then, skin lesions are segmented by TransUNet, which is tuned using artificial ecosystem‐based optimization (AEO). Next, the segmented image is augmented, and then the features involving Haralick textural, statistical, convolutional neural network (CNN), and shape local binary texture (SLBT) features are extracted. Following this, skin cancer detection is performed by utilizing LeNet. Here, LeNet is tuned by Jaya artificial ecosystem‐based optimization (JAEO), which is formed by combining AEO and Jaya optimization. Next, skin cancer classification is executed using TANF‐ShCNN. This TANF‐ShCNN method is produced by fusing the adaptive neuro‐fuzzy inference system (ANFIS) and ShCNN based on the Taylor concept. Moreover, the classification efficiency of TANF‐ShCNN measured a maximum accuracy of 92.535%, specificity of 92.80%, and sensitivity of 93.48%.