No abstract

Breast Cancer Classification is important in the medical field for disease diagnosis and assists in decisions in treatment. Poor convergence and local optima are common limitations in the existing feature selection techniques. Overfitting and imbalance data problems are common limitations in existing classifiers. The hybrid method of Whale Optimization Algorithm (WOA) – Slime Mould Algorithm (SMA) is proposed for relevant feature selection and an Auto stacked encoder is applied for classification. The WOA technique performs exploration in the first half of the iteration and the SMA method performs exploitation in the second half of iterations for relevant feature selection. The WOA-SMA technique is applied in BreakHis and IDC datasets to evaluate breast cancer classification. The ResNet18 model uses the convolutional layer, pooling layer, and fully connected layer for feature extraction process. The stacked autoencoder technique helps to map relevant features to increase learning rate of the model. The feature selection method of WOA-SMA selects unique features to improve learning efficiency of the model.

No abstract