A Mathematical Approach Towards Quantization of Floating Point Weights in Low Power Neural Networks | IEEE Conference Publication | IEEE Xplore