Used CNN layers used TensorFlow keras library Sequential->convolutional->pooling(avg pooling)->flatten for feed forward Used 'adam' as optimizer and sparse categorical cross entropy as loss function for back propagation Evaluated using accuracy and compare the plots using the matplotlib library of python between train and test set