Published Paper Details:

Antibody microarrays in a patent-pending series may be initiated and validated using microarray technology. A Microarray Data Analysis (MDA) is used to determine the patterns of hundreds of genes within a single experiment. There is a vast amount of gene expression data in the MD that may be used to diagnose malignancy. However, over-fitting and under-fitting issues arise due to the unbalanced class label instances present in microarray gene datasets and the initialization parameter value for the classifier. To get over this obstacle, this study proposes a stacking ensemble of Deep cluster-based DL systems for Cancer Classification. This system combines many learning models into a single, highly accurate prediction model. There are three distinct parts to the created model. To begin, we create a Modified Harmony Search Algorithm and a Modified Kernel-based Fuzzy C-Means (MHSAMKFC) to efficiently remove massive duplicate features. Second, to deal with uncertainties in the labeled training dataset and boost classifier performance, the MHSAMKFC with Convolutional Neural Network (CNN) classifier is suggested. Third, the ensemble technique, which employs several learning models to improve prediction accuracy, mitigates MHSAMKF0C over-fitting CNNs and under-fitting issues. En-MHSAMKFC-CNN describes the whole operation. In a conclusion, experiments are run on four Gene Expression Microarray (GEM) datasets to confirm that the En-MHSAMKFC-CNN enhances the classification performance of SVM, KNN, RF, and ANN classifiers.