Genetically Optimized Deep Neural Learning for Breast Cancer Prediction

Abstract

Breast cancer is one of the most commonly occurring cancers among women. Early detection of the disease is quite tricky as it involves multiple numbers of tests to evaluate factors such as location, type and size of tumor which influence the valid identification of cancer that can be time-consuming for the doctor. To increase the efficiency of disease detection, a deep neural network with an evolutionary optimization technique is implemented which will target at searching optimal weights simultaneously for multiple neurons of the net. It involves the optimization of weights and the number of iterations required to train the machine which will help in better prediction of cancer. This is experimented to classify the breast cancer (Wisconsin Breast Cancer Dataset) into benign and malignant classes. This technique will help in reducing the time required to diagnose breast cancer at an early stage making treatment effective. The experimentation result is comparable with other state-of-the-art methods in terms of the classification accuracy. © 2020, Springer Nature Singapore Pte Ltd.