During the training process, the neural network learns to perform the expected task. For every iteration/epoch, the neural network will evaluate its training knowledge. Accordingly, it will re-iterate the layers with updated gradient values to minimize the error produced at the output layer. Also, note that labels (0 and 1 ) are not uniformly distributed across the dataset. So, we might need to consider adding weights to the label that appears less in the dataset. This is highly recommended before we proceed with the actual training session. In this recipe, we will train the neural network and evaluate the resultant model.