Machine Learning for Broadband Complex Permittivity Based Accurate Detection Technology

A novel machine learning based method over a wide-band for the detection of the temperature and concentration of water solution is provided. Compared with the traditional interpolation fitting methods, like the Debye equation, wide-band prediction can avoid the contingency of single frequency point and increase the prediction accuracy. Moreover, this method does note need to fit the equation, so it is easier to implement and use. After using measured data for test, the XGBooster can catch the highest accuracy compared to the traditional methods and far exceeding that of the interpolation method, and meanwhile obtain the least time complexity.