Breast Cancer Classification Using Ensemble Approach, Machine Learning and Deep Learning
N Kushal Kumar Raju, Avinash Khatua, S Tarun, Mohan Monica Subashini · 2022
This research aims to use data classification, image classification, and the ensemble approach to classify and predict benign or malignant cancer cells, therefore lowering the time and expense involved with breast cancer screening. The dataset used in data classification was augmented using the Data Augmentation approach, and the augmented data is then separated into train, test, and validation. The model is then trained using CNN, SVM, KNN, and Random Forest algorithms, after which it is verified with the test data and validation parts of the dataset. The image classification models are trained using CNN and Random Forest algorithms then the models are validated by predicting images of different dataset image to check if a cell is benign or malignant. In data classification after 50 epochs of training, an accuracy of 94.73 percent was achieved for CNN in categorizing data, 92.98 percent for SVM, 90.87 percent for KNN, and 95.3 percent for Random Forest, in image classification 92.36 percent for CNN and 93.28 percent for Random Forest are achieved.