Reduced precision discretization based on information theory

In recent years, new technological areas have emerged and proliferated, such as the Internet of Things or embedded systems in drones, which are usually characterized by making use devices with strict requirements weight, size, cost power consumption. As a consequence, there has been growing interest implementation machine learning algorithms reduced precision that can be these constrained devices. These cover not only learning, but they also applied to other stages feature selection data discretization. this work we study behavior Minimum Description Length Principle (MDLP) discretizer, proposed Fayyad Irani, when is used, how much it affects typical pipeline. Experimental results show fixed-point format sufficient achieve performances similar those obtained using double-precision format, opens door reduced-precision discretizers systems, minimizing energy consumption carbon emissions.