Selection of Gradient Descent Optimizers for Convolutional Neural Network Based Brain Tumor Detectors

These days, brain tumor discovery has turned up as an overall causality in the domain of medical care. The venture for tumor location begins with the procurement of MRI check image of tumor. This research paper proposes a programmed division strategy that depends upon Convolution Neural Networks (CNN). By consolidating this single method, classification is cultivated. In this paper the efficiency of various gradient descent optimizers on the proposed brain tumor detector model has been tested. In order to verify the efficiency of the optimizers, a loss function based on categorical cross entropy is defined. Next, many gradient descent optimizers such as Stochastic Gradient Descent (SGD), AdaMax, Adadelta, AdaGrad, and Nadam are selected. These optimizers are applied to train and validate the CNN model using uniform values of learning rate parameters and number of epochs. As a result of comparative study, the Adadelta optimizer has provided least value of loss function.