Towards optimal quantization of neural networks | IEEE Conference Publication | IEEE Xplore