Comparison of Floating-point Representations for the Efficient Implementation of Machine Learning Algorithms | IEEE Conference Publication | IEEE Xplore