Breast Cancer Detection Using Novel Weight Based Convolutional Neural Network With Simulated Annealing (Nwcnn-Sa) Based Outlier Detection Algorithm Combined With Weighted Deep Autoencoder (Wdae) - Chimp Optimization Algorithm (Choa) For Classification

Abstract

One of the main causes of mortality worldwide, which has been rising in recent years, is cancer. There are two categories for this illness: benign and malignant. Breast cancer, which manifests in breast tissue despite a person's gender, is a single of the initial and largest types of cancer in the human body. The detection of cancer of the breast has benefited greatly from the application of deep learning. This study applies the selection of features and modification to the Wisconsin Diagnostic Breast Cancer dataset. Finding the most pertinent traits is meant to help categorize a diagnosis as benign or malignant. In the beginning of this study project, the data's dimensionality is reduced using the Non-Negative Matrix Factorization (NMF) Algorithm. After that, the outliers from the cancer dataset are found using the Novel weight-based Convolutional neural network with Simulated Annealing (NWCNN-SA) detection approach. Then, Remora Optimization Algorithm (ROA) is used to pick the attributesfrom the data with the outliers eliminated. After that, all of the altered images are given to the Weighted Deep AutoEncoder (WDAE) with Chimp Optimization Algorithm (ChOA), also known as the WADE-ChOAclassifier, for the training process. This helps to maximize the detection accuracy by classifying incoming clinical images of breast cancer as benign or malignant without requiring any prior knowledge about the appearance of cancer. Utilizing the breast cancer database, evaluations the efficacy of the suggested technique. A dataset of breast cancer cases, prior to as well as following the elimination of outliers, was used for the experiments. The results show that the proposed strategy is more accurate and performs better. The findings of the research on breast cancer will help in patient diagnosis and care. The suggested method's effectiveness has been evaluated using the breast cancer database. The trials employed a dataset of instances of breast cancer, both prior to and following the outliers were removed. The outcomes demonstrate how much more precise and effective the suggested method is. The results of this breast cancer study might help in detection and supervision of patients.