Implementation of optimization algorithms on Wisconsin Breast cancer dataset using deep neural network

Abstract

In women, the 2nd most common cancer is breast cancer annually about 16 lakh women are identified. Detection and treatment in the early stage improve survival rates. Dataset of Wisconsin Breast cancer will provide the features of the digitized image. In this paper, a model is developed and different optimization algorithms were implemented to access the correctness of classifying data with respect to accuracy which is feasible for computer-aided diagnosis. Machine learning can assist and alert expert radiologist more effectively than current screening techniques. In this paper RMS propagation (Root Mean Square Propagation) and SGD (stochastic gradient descent) optimization algorithms were implemented on a deep neural network with sigmoid neurons and accuracy is compared.