China population prediction with the aid of fast, optimal and accurate WASD neuronet compared with BP neuronet

Population is one of the most significant factors affecting the social and economic development of a country. Currently, China is facing a big challenge of population aging. It is important for Chinese government to pay continuous attention to the population situation and take measures so as to keep the economic growth. Undoubtedly, the effectiveness of these measures depends crucially on the accurate prediction of China population. Many methods, including the feedforward neuronet equipped with back-propagation (BP) algorithm, have been proposed in order to improve the prediction accuracy. However, these methods possibly have their inherent limitations. In this paper, the authors construct an enhanced feedforward neuronet, i.e., a 3-layer feedforward power-activated neuronet (PAN) equipped with the weights-and-structure-determination algorithm (for short, WASD-PAN), to predict China population. Numerical experiments show that China population will continue to increase at a low rate in the next decade. The authors also compare the WASD-PAN with the back-propagation neuronet BP-PAN, with numerical experimental results showing the excellent performance of the former.