Using Multilayer Perceptron and Deep Neural Networks for the Diagnosis of Breast Cancer Classification

Abstract

The very long delay that is suffered by patients of breast cancer in their early stages in low-income countries is due to access barriers and quality deficiencies in the care of cancer giving rise to the need for an alternative and efficient computer-based diagnostic system for the early detection and prevention of the disease. The early detection and improved therapy still remain a crucial approach for the prevention and cure of breast cancer. To this end, recent research looks into the development of different classifier models for the classification of breast cancer. This paper investigates the potentials of applying multiple neural network architectures with increased number of hidden layers and hidden units. The network architectures have one-hidden-layer, two-hidden-layer and three hidden layer (deep neural network) architectures respectively using the backpropagation training algorithm for the training of the models. The experimental results show that by applying this approach the models yield efficient and promising results