In this paper, we mainly study a new method for estimating the Mg abundance of stars from the medium and low resolution spectra based on ELM algorithm. LAMOST provides us with a massive low-resolution spectrum, and determining the abundance of Mg elements in these spectra will help us understand the history and evolution of the Milky Way. At present, the traditional method of determining the abundance of Mg element from medium and low resolution spectra is the template matching method. However, this method is difficult to optimize parameters and is sensitive to noise. Therefore, it is necessary to study new methods to estimate the Mg abundance. The experiment show that ELM algorithm is agood alternative to traditional method. The accuracy of ELM algorithm on MILES spectra is 0.009 9 (0.15) dex, while on the LAMOST spectra with signal-to-noise ratios larger than 50 it is 0.002 7 (0.11) dex. A comparison of ELM with other four algorithms shows that ELM algorithm can accurately estimate the abundance of Mg elements from low resolution spectra and can be applied to the LAMOST spectra.