CHaPR: Efficient Inference of CNNs via Channel Pruning | IEEE Conference Publication | IEEE Xplore