Sensitivity-based acceleration and compression algorithm for convolution neural network | IEEE Conference Publication | IEEE Xplore