Functional Network for Nonlinear Regression Based on Extreme Learning Machine

Abstract

Functional network is an extension of neural network proposed in recent years, which is becoming a research hotspot in the field of machine learning. Similar to neural network, the learning algorithm of functional network also uses the parameter iteration method. However, one of the disadvantages of this method is time-consuming, seriously affecting the learning efficiency of our network. In order to solve this problem, in this paper, we proposed a new functional network model for solving the nonlinear regression forecast problems, termed as functional network for nonlinear regression based on extreme learning machine (FN-ELM). In FN-ELM, the idea of extreme learning machine (ELM) is introduced into functional network, making the whole learning process of functional network without iteration. This method solves the time-consuming problem of the traditional functional network very well. Simulation results show that our method can improve the forecasting accuracy and learning efficiency compared with ELM and functional network.