Loss Constrains Added Squeeze and Excitation Blocks for Pruning Deep Neural Networks | IEEE Conference Publication | IEEE Xplore