Abstract: We implement transfer learning (TL) to attain efficiency in training and feature extraction by freezing the connection weights of shallow layers of deep learning (DL) models, already trained ...