Reduce Computing Complexity of Deep Neural Networks Through Weight Scaling | IEEE Conference Publication | IEEE Xplore