Advanced Machine Learning Techniques for Early Detection and Classification of Breast Cancer

Abstract

Breast cancer is one of the most dangerous diseases, and it is the second leading cause of morbidity in women. Breast cancer arises when hazardous, malignant tumours grow in the mammary gland. Self-tests and routine clinical checks help in early detection and hence substantially improve survivorship. Breast cancer classification is a medical method that academics and specialists find difficult to implement. Several microarray studies have utilised gene signatures to create classifications that predict medical outcomes for various cancer patients. Signatures from diverse studies usually suffer from low consistency when used in the classification of databases, regardless of the study from which they were produced.  By integrating the auto-encoder and Principal Component Analysis, the researchers provide an unsupervised feature training strategy for characterizing different qualities from variations in gene expression. An ensemble classifier based on the AdaBoost algorithm was created as the framework for the gathered attributes to anticipate medical outcomes in breast cancer. During the experiments, the researchers created an additional classifier using the same classifier learning strategy to act as a median for the suggested technique. Experiments reveal that the proposed system, which makes use of deep learning techniques, outperforms others.