A new device characteristic model generation by machine learning

This paper proposes a method to generate a characteristic model of a new device by machine learning. It is difficult new emerging devices and materials to have exact physical model to fit the behavior because developing physical models is complex and takes a long time. Therefore, we propose a regression method by machine learning as an alternative intermediate step of the physical model, and present guidelines for date preprocessing and proper architecture for efficient learning. It makes it easier and faster to solve general regression problems such as function generation.