Design and implementation of low-level machine learning API and API server

The purpose of this paper is to understand various problems related to Computer Science and find solutions from optimized machine learning algorithm by realizing machine learning API and API server. The representative machine-learning algorithm, TensorFlow, need to express algorithm from the stage of nodes and edges while IBM Watson only uses functions in completed form. Those are the problems to be solved in this paper; therefore, we suggest low-level API, which is in the middle stage between TensorFlow and IBM Watson and realize the low-level API. The machine learning API is realized in machine learning of artificial neural network and stochastic shunt, which will be applicable in other areas.