Research on Extreme Learning Machine Algorithm and Its Application to El-Niño/La-Niña Southern Oscillation Model

Since it has the ability to give a faster result than traditional machine learning algorithms, Extreme Learning Machine (ELM) has become increasingly popular in various research fields recently. However, ELM has been worked on the research of computer science and other related areas except for the atmospheric field. This paper uses the ELM algorithm to simulate the forward integrating process of an ocean-atmosphere oscillator model – El-Niño/La-Niña Southern Oscillation (ENSO). The results show that the ELM algorithm has a good accuracy and efficiency with a quick convergence speed and a strong resistance over observation noises.